catherd-cli 1.4.0 → 2.0.0-beta.20261022113309

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (320) hide show
  1. package/README.md +188 -169
  2. package/THIRD_PARTY_NOTICES.md +217 -35
  3. package/catalog/jev.json +27 -1
  4. package/catalog/models.json +56 -239
  5. package/package.json +6 -8
  6. package/plugin/.claude-plugin/plugin.json +1 -1
  7. package/plugin/.codex-plugin/plugin.json +1 -1
  8. package/plugin/bin/catherd-mcp +1 -1
  9. package/plugin/commands/catherd.md +2 -2
  10. package/plugin/skills/catherd/SKILL.md +97 -356
  11. package/plugin/skills/catherd-setup/SKILL.md +21 -20
  12. package/src/adapters/all.ts +0 -4
  13. package/src/adapters/backend.ts +19 -4
  14. package/src/adapters/claude-code/index.ts +126 -14
  15. package/src/adapters/codex/events.ts +10 -0
  16. package/src/adapters/codex/index.ts +17 -5
  17. package/src/adapters/cursor/home.ts +1 -1
  18. package/src/adapters/cursor/index.ts +10 -4
  19. package/src/adapters/opencode/index.ts +38 -1
  20. package/src/cli.ts +27 -27
  21. package/src/domain/agents.ts +3 -1
  22. package/src/domain/bars.ts +33 -107
  23. package/src/domain/blocked.ts +72 -0
  24. package/src/domain/branch.ts +51 -0
  25. package/src/domain/brief.ts +106 -0
  26. package/src/domain/catalog.ts +69 -313
  27. package/src/domain/changes.ts +23 -1
  28. package/src/domain/cost.ts +19 -29
  29. package/src/domain/errors.ts +6 -2
  30. package/src/domain/fail-lines.ts +26 -0
  31. package/src/domain/failover.ts +37 -75
  32. package/src/domain/findings.ts +47 -0
  33. package/src/domain/gate-brief.ts +53 -0
  34. package/src/domain/hints.ts +13 -2
  35. package/src/domain/ids.ts +13 -2
  36. package/src/domain/jev.ts +47 -2
  37. package/src/domain/judge.ts +124 -0
  38. package/src/domain/lane.ts +118 -3
  39. package/src/domain/milestone-memory.ts +53 -0
  40. package/src/domain/notice.ts +32 -8
  41. package/src/domain/notify.ts +32 -0
  42. package/src/domain/plan-doc.ts +198 -0
  43. package/src/domain/plan-lint.ts +141 -0
  44. package/src/domain/profile-rules.ts +175 -138
  45. package/src/domain/profile.ts +140 -63
  46. package/src/domain/record.ts +68 -3
  47. package/src/domain/reply.ts +29 -0
  48. package/src/domain/role-prompts.ts +182 -56
  49. package/src/domain/role-scope.ts +76 -0
  50. package/src/domain/role-tools.ts +34 -0
  51. package/src/domain/roles.ts +2 -0
  52. package/src/domain/route.ts +90 -0
  53. package/src/domain/rules.ts +33 -0
  54. package/src/domain/run-events.ts +298 -0
  55. package/src/domain/run-machine.ts +668 -0
  56. package/src/domain/run-state.ts +1310 -0
  57. package/src/domain/scorecard.ts +197 -0
  58. package/src/domain/select.ts +229 -72
  59. package/src/domain/slopalytics.ts +72 -0
  60. package/src/domain/state.ts +5 -0
  61. package/src/domain/util.ts +22 -0
  62. package/src/domain/workspace.ts +130 -0
  63. package/src/entry/answer-command.ts +39 -0
  64. package/src/entry/catalog-command.ts +24 -165
  65. package/src/entry/check-bin.ts +17 -0
  66. package/src/entry/cli-kit.ts +18 -2
  67. package/src/entry/daemon-bin.ts +22 -0
  68. package/src/entry/deps.ts +2 -0
  69. package/src/entry/doctor-command.ts +20 -2
  70. package/src/entry/glyphs.ts +15 -2
  71. package/src/entry/host-arg.ts +28 -0
  72. package/src/entry/init-command.ts +20 -69
  73. package/src/entry/knowledge-command.ts +121 -5
  74. package/src/entry/lock-command.ts +103 -8
  75. package/src/entry/mcp/handshake.ts +6 -2
  76. package/src/entry/mcp/role-bin.ts +20 -0
  77. package/src/entry/mcp/role-server.ts +134 -0
  78. package/src/entry/mcp/server.ts +85 -21
  79. package/src/entry/mcp/session-tools.ts +149 -0
  80. package/src/entry/panel/openapi.ts +189 -0
  81. package/src/entry/panel/server.ts +410 -0
  82. package/src/entry/panel-command.ts +68 -0
  83. package/src/entry/pause-command.ts +67 -0
  84. package/src/entry/profile-command.ts +33 -18
  85. package/src/entry/role-cli-command.ts +159 -0
  86. package/src/entry/runs-command.ts +81 -27
  87. package/src/entry/scorecard-command.ts +187 -0
  88. package/src/entry/supervise-bin.ts +1 -1
  89. package/src/entry/supervise-command.ts +30 -3
  90. package/src/entry/workspace-command.ts +44 -0
  91. package/src/infra/boot-lock.ts +25 -0
  92. package/src/infra/delivery.ts +3 -0
  93. package/src/infra/dispatch-dir.ts +2 -0
  94. package/src/infra/env.ts +11 -3
  95. package/src/infra/git.ts +70 -0
  96. package/src/infra/heavy-lock.ts +133 -6
  97. package/src/infra/host-probe.ts +16 -0
  98. package/src/infra/launch.ts +63 -7
  99. package/src/infra/lock-activity.ts +88 -0
  100. package/src/infra/login-env.ts +48 -0
  101. package/src/infra/memory.ts +45 -0
  102. package/src/infra/ntfy.ts +40 -0
  103. package/src/infra/open-url.ts +42 -0
  104. package/src/infra/paths.ts +37 -2
  105. package/src/infra/role-mcp.ts +69 -0
  106. package/src/infra/slopalytics.ts +30 -0
  107. package/src/infra/store.ts +6 -1
  108. package/src/infra/supervisor.ts +107 -16
  109. package/src/services/admission.ts +133 -31
  110. package/src/services/backends.ts +4 -6
  111. package/src/services/budget.ts +26 -6
  112. package/src/services/capture.ts +1 -22
  113. package/src/services/catalog-service.ts +155 -254
  114. package/src/services/credentials.ts +15 -26
  115. package/src/services/daemon.ts +181 -0
  116. package/src/services/dispatch-service.ts +209 -17
  117. package/src/services/dispatches.ts +30 -2
  118. package/src/services/doctor-access.ts +8 -3
  119. package/src/services/doctor-backends.ts +1 -1
  120. package/src/services/doctor-checks.ts +47 -0
  121. package/src/services/doctor-docker.ts +221 -0
  122. package/src/services/doctor-role-mcp.ts +65 -0
  123. package/src/services/doctor.ts +72 -15
  124. package/src/services/end-push.ts +157 -0
  125. package/src/services/failover-rungs.ts +109 -0
  126. package/src/services/finalize.ts +24 -2
  127. package/src/services/gate-env.ts +217 -0
  128. package/src/services/gate-service.ts +231 -45
  129. package/src/services/global-install.ts +46 -13
  130. package/src/services/jev-service.ts +2 -4
  131. package/src/services/lane-edit.ts +154 -0
  132. package/src/services/lane-service.ts +291 -55
  133. package/src/services/milestone-check.ts +433 -0
  134. package/src/services/milestone-tree.ts +170 -0
  135. package/src/services/milestones.ts +167 -29
  136. package/src/services/notifier.ts +124 -4
  137. package/src/services/notify-service.ts +26 -0
  138. package/src/services/orchestrator-wait.ts +82 -0
  139. package/src/services/panel-setup.ts +116 -0
  140. package/src/services/pause.ts +161 -0
  141. package/src/services/peek.ts +79 -17
  142. package/src/services/plan-input.ts +151 -0
  143. package/src/services/ports.ts +60 -5
  144. package/src/services/preflight.ts +232 -24
  145. package/src/services/profile-service.ts +31 -3
  146. package/src/services/profile-store.ts +18 -6
  147. package/src/services/protocol.ts +123 -25
  148. package/src/services/provenance.ts +34 -89
  149. package/src/services/questions.ts +24 -0
  150. package/src/services/reconcile.ts +33 -1
  151. package/src/services/reentry.ts +7 -6
  152. package/src/services/role-access.ts +27 -0
  153. package/src/services/routing-service.ts +161 -42
  154. package/src/services/run-briefs.ts +477 -0
  155. package/src/services/run-client.ts +217 -0
  156. package/src/services/run-effects.ts +837 -0
  157. package/src/services/run-events-store.ts +174 -0
  158. package/src/services/run-evidence.ts +1 -1
  159. package/src/services/run-pin.ts +125 -0
  160. package/src/services/run-read.ts +221 -0
  161. package/src/services/run-report.ts +176 -0
  162. package/src/services/run-service.ts +204 -23
  163. package/src/services/run-store.ts +159 -18
  164. package/src/services/scorecard-metrics.ts +86 -0
  165. package/src/services/scorecard.ts +382 -0
  166. package/src/services/scratch.ts +60 -0
  167. package/src/services/sessions.ts +3 -1
  168. package/src/services/state.ts +21 -0
  169. package/src/services/summary.ts +76 -12
  170. package/src/services/tree-path.ts +27 -0
  171. package/src/services/verifier-brief.ts +23 -0
  172. package/src/services/verifier-step.ts +48 -0
  173. package/src/services/workspace-admission.ts +259 -0
  174. package/src/services/workspace-service.ts +259 -0
  175. package/src/services/workspace-store.ts +95 -0
  176. package/web/dist/assets/Brand-PKlUX5TE.js +1 -0
  177. package/web/dist/assets/Brand-PKlUX5TE.js.br +0 -0
  178. package/web/dist/assets/Brand-PKlUX5TE.js.gz +0 -0
  179. package/web/dist/assets/Modal-Dxc5so53.js +1 -0
  180. package/web/dist/assets/Modal-Dxc5so53.js.br +0 -0
  181. package/web/dist/assets/Modal-Dxc5so53.js.gz +0 -0
  182. package/web/dist/assets/Pressable-C8-T6Ed_.js +1 -0
  183. package/web/dist/assets/StatusChip-B5xfI_Mn.js +153 -0
  184. package/web/dist/assets/StatusChip-B5xfI_Mn.js.br +0 -0
  185. package/web/dist/assets/StatusChip-B5xfI_Mn.js.gz +0 -0
  186. package/web/dist/assets/_app-Ea5cpyKe.js +1 -0
  187. package/web/dist/assets/_app-Ea5cpyKe.js.br +0 -0
  188. package/web/dist/assets/_app-Ea5cpyKe.js.gz +0 -0
  189. package/web/dist/assets/_runId-pTG0fXVo.js +1 -0
  190. package/web/dist/assets/_runId-pTG0fXVo.js.br +0 -0
  191. package/web/dist/assets/_runId-pTG0fXVo.js.gz +0 -0
  192. package/web/dist/assets/drawer-Du3cdlok.js +1 -0
  193. package/web/dist/assets/drawer-Du3cdlok.js.br +0 -0
  194. package/web/dist/assets/drawer-Du3cdlok.js.gz +0 -0
  195. package/web/dist/assets/geist-cyrillic-ext-wght-normal-DjL33-gN.woff2 +0 -0
  196. package/web/dist/assets/geist-cyrillic-wght-normal-BEAKL7Jp.woff2 +0 -0
  197. package/web/dist/assets/geist-latin-ext-wght-normal-DC-KSUi6.woff2 +0 -0
  198. package/web/dist/assets/geist-latin-wght-normal-BgDaEnEv.woff2 +0 -0
  199. package/web/dist/assets/geist-mono-cyrillic-ext-wght-normal-X_5orZeX.woff2 +0 -0
  200. package/web/dist/assets/geist-mono-cyrillic-wght-normal-DiZS0aHC.woff2 +0 -0
  201. package/web/dist/assets/geist-mono-latin-ext-wght-normal-Bwz-egvJ.woff2 +0 -0
  202. package/web/dist/assets/geist-mono-latin-wght-normal-XN7g48iV.woff2 +0 -0
  203. package/web/dist/assets/geist-mono-symbols2-wght-normal-CO5SzqOn.woff2 +0 -0
  204. package/web/dist/assets/geist-mono-vietnamese-wght-normal-DadHysG0.woff2 +0 -0
  205. package/web/dist/assets/geist-vietnamese-wght-normal-6IgcOCM7.woff2 +0 -0
  206. package/web/dist/assets/index-B6GiwDkh.js +11 -0
  207. package/web/dist/assets/index-B6GiwDkh.js.br +0 -0
  208. package/web/dist/assets/index-B6GiwDkh.js.gz +0 -0
  209. package/web/dist/assets/index-C5IbylvM.css +2 -0
  210. package/web/dist/assets/index-C5IbylvM.css.br +0 -0
  211. package/web/dist/assets/index-C5IbylvM.css.gz +0 -0
  212. package/web/dist/assets/label-DFSYBc62.js +1 -0
  213. package/web/dist/assets/label-DFSYBc62.js.br +0 -0
  214. package/web/dist/assets/label-DFSYBc62.js.gz +0 -0
  215. package/web/dist/assets/link-D8EcKZUt.js +1 -0
  216. package/web/dist/assets/link-D8EcKZUt.js.br +0 -0
  217. package/web/dist/assets/link-D8EcKZUt.js.gz +0 -0
  218. package/web/dist/assets/login-DaN5HPPW.js +1 -0
  219. package/web/dist/assets/login-DaN5HPPW.js.br +0 -0
  220. package/web/dist/assets/login-DaN5HPPW.js.gz +0 -0
  221. package/web/dist/assets/query-Cu--ccT3.js +1 -0
  222. package/web/dist/assets/questions-shoO9OVj.js +1 -0
  223. package/web/dist/assets/questions-shoO9OVj.js.br +0 -0
  224. package/web/dist/assets/questions-shoO9OVj.js.gz +0 -0
  225. package/web/dist/assets/react-query.gen-Bu2yV-qJ.js +12 -0
  226. package/web/dist/assets/react-query.gen-Bu2yV-qJ.js.br +0 -0
  227. package/web/dist/assets/react-query.gen-Bu2yV-qJ.js.gz +0 -0
  228. package/web/dist/assets/rolldown-runtime-XKyaHtl9.js +1 -0
  229. package/web/dist/assets/run-status-C0DRuV7U.js +1 -0
  230. package/web/dist/assets/runs-C3rg2zZc.js +1 -0
  231. package/web/dist/assets/runs-C3rg2zZc.js.br +0 -0
  232. package/web/dist/assets/runs-C3rg2zZc.js.gz +0 -0
  233. package/web/dist/assets/setup-5I4Vl9fR.js +1 -0
  234. package/web/dist/assets/setup-5I4Vl9fR.js.br +0 -0
  235. package/web/dist/assets/setup-5I4Vl9fR.js.gz +0 -0
  236. package/web/dist/assets/table-BOxXhqS4.js +1 -0
  237. package/web/dist/assets/table-BOxXhqS4.js.br +0 -0
  238. package/web/dist/assets/table-BOxXhqS4.js.gz +0 -0
  239. package/web/dist/assets/textfield-BzRFkxBF.js +1 -0
  240. package/web/dist/assets/textfield-BzRFkxBF.js.br +0 -0
  241. package/web/dist/assets/textfield-BzRFkxBF.js.gz +0 -0
  242. package/web/dist/assets/time--9kSQbbp.js +1 -0
  243. package/web/dist/assets/use-css-variable-CpnN9WtA.js +5 -0
  244. package/web/dist/assets/use-css-variable-CpnN9WtA.js.br +0 -0
  245. package/web/dist/assets/use-css-variable-CpnN9WtA.js.gz +0 -0
  246. package/web/dist/assets/useMutation-CVPKlR34.js +1 -0
  247. package/web/dist/assets/useMutation-CVPKlR34.js.br +0 -0
  248. package/web/dist/assets/useMutation-CVPKlR34.js.gz +0 -0
  249. package/web/dist/index.html +15 -0
  250. package/catalog/ATTRIBUTION.md +0 -36
  251. package/catalog/scores.json +0 -3202
  252. package/catalog/sources.json +0 -97
  253. package/src/adapters/antigravity/events.ts +0 -131
  254. package/src/adapters/antigravity/home.ts +0 -46
  255. package/src/adapters/antigravity/index.ts +0 -320
  256. package/src/adapters/antigravity/models.ts +0 -37
  257. package/src/adapters/grok/events.ts +0 -112
  258. package/src/adapters/grok/home.ts +0 -132
  259. package/src/adapters/grok/index.ts +0 -339
  260. package/src/adapters/grok/models.ts +0 -48
  261. package/src/domain/calibration.ts +0 -125
  262. package/src/domain/sources.ts +0 -144
  263. package/src/entry/mcp/dispatch-tools.ts +0 -52
  264. package/src/entry/mcp/lane-tools.ts +0 -84
  265. package/src/entry/mcp/protocol-tools.ts +0 -57
  266. package/src/entry/mcp/run-tools.ts +0 -148
  267. package/src/entry/mcp/setup-tools.ts +0 -108
  268. package/src/entry/tui/commands.ts +0 -563
  269. package/src/entry/tui/effects.ts +0 -381
  270. package/src/entry/tui/fixtures.ts +0 -637
  271. package/src/entry/tui/profile-edits.ts +0 -146
  272. package/src/entry/tui/profile-tree.ts +0 -429
  273. package/src/entry/tui/providers/app.tsx +0 -162
  274. package/src/entry/tui/providers/boundary.tsx +0 -57
  275. package/src/entry/tui/providers/data.tsx +0 -160
  276. package/src/entry/tui/providers/keymap.tsx +0 -166
  277. package/src/entry/tui/providers/root.tsx +0 -45
  278. package/src/entry/tui/providers/theme.tsx +0 -17
  279. package/src/entry/tui/providers/toast.tsx +0 -77
  280. package/src/entry/tui/run.tsx +0 -114
  281. package/src/entry/tui/state.ts +0 -373
  282. package/src/entry/tui/stories.ts +0 -114
  283. package/src/entry/tui/text.ts +0 -130
  284. package/src/entry/tui/theme.ts +0 -140
  285. package/src/entry/tui/views/app.tsx +0 -208
  286. package/src/entry/tui/views/dialogs.tsx +0 -41
  287. package/src/entry/tui/views/profile-actions.ts +0 -331
  288. package/src/entry/tui/views/profiles.tsx +0 -520
  289. package/src/entry/tui/views/runs.tsx +0 -573
  290. package/src/entry/tui/views/save-dialog.tsx +0 -340
  291. package/src/entry/tui/views/status.tsx +0 -209
  292. package/src/entry/tui/widgets/chrome.tsx +0 -114
  293. package/src/entry/tui/widgets/dialog-confirm.tsx +0 -58
  294. package/src/entry/tui/widgets/dialog-prompt.tsx +0 -52
  295. package/src/entry/tui/widgets/dialog-select.tsx +0 -212
  296. package/src/entry/tui/widgets/dialog.tsx +0 -68
  297. package/src/entry/tui/widgets/line.tsx +0 -92
  298. package/src/entry/tui/widgets/list.tsx +0 -197
  299. package/src/entry/tui/widgets/spinner.tsx +0 -49
  300. package/src/entry/tui/widgets/toast.tsx +0 -62
  301. package/src/infra/sources/arena.ts +0 -87
  302. package/src/infra/sources/artificial-analysis.ts +0 -112
  303. package/src/infra/sources/cache.ts +0 -124
  304. package/src/infra/sources/csv.ts +0 -38
  305. package/src/infra/sources/epoch.ts +0 -82
  306. package/src/infra/sources/http.ts +0 -75
  307. package/src/infra/sources/litellm.ts +0 -58
  308. package/src/infra/sources/models-dev.ts +0 -59
  309. package/src/infra/sources/openrouter-endpoints.ts +0 -52
  310. package/src/infra/sources/openrouter-models.ts +0 -60
  311. package/src/infra/sources/rows.ts +0 -38
  312. package/src/infra/sources/vectara.ts +0 -69
  313. package/src/infra/sources/zip.ts +0 -48
  314. package/src/services/catalog-refresh.ts +0 -201
  315. package/src/services/doctor-sources.ts +0 -76
  316. package/src/services/runs-page.ts +0 -255
  317. package/src/services/source-derive.ts +0 -374
  318. package/src/services/source-sync.ts +0 -305
  319. package/src/services/standins.ts +0 -204
  320. package/src/services/treat-likes.ts +0 -97
package/catalog/scores.json DELETED
@@ -1,3202 +0,0 @@
1
- {
2
- "schema": 1,
3
- "version": "2026-09-28",
4
- "benchmarks": {
5
- "repo_code": {
6
- "benchmark": "DeepSWE",
7
- "version": "1.1"
8
- },
9
- "terminal": {
10
- "benchmark": "Terminal-Bench",
11
- "version": "4.0"
12
- },
13
- "honesty": {
14
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
15
- "version": "GPT-6 system card appendix 11.6.4.2"
16
- },
17
- "agentic": {
18
- "benchmark": "Arena agent, net improvement",
19
- "version": "lmarena-ai/leaderboard-dataset"
20
- },
21
- "steer": {
22
- "benchmark": "Arena agent steerability",
23
- "version": "lmarena-ai/leaderboard-dataset"
24
- },
25
- "frontend": {
26
- "benchmark": "Arena WebDev rating",
27
- "version": "lmarena-ai/leaderboard-dataset"
28
- }
29
- },
30
- "scores": [
31
- {
32
- "rung": "claude-fable-5-1#high",
33
- "dim": "terminal",
34
- "value": 55.8,
35
- "benchmark": "Terminal-Bench",
36
- "version": "4.0",
37
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
38
- "date": "2026-09-22",
39
- "confidence": "adjacent",
40
- "note": "Terminal-Bench has it at max; carried to this effort"
41
- },
42
- {
43
- "rung": "claude-fable-5-1#high",
44
- "dim": "agentic",
45
- "value": 0.1384,
46
- "benchmark": "Arena agent, net improvement",
47
- "version": "2026-09-27",
48
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
49
- "date": "2026-09-27",
50
- "confidence": "adjacent",
51
- "source": "arena",
52
- "note": "arena has it at max; carried to this effort"
53
- },
54
- {
55
- "rung": "claude-fable-5-1#high",
56
- "dim": "steer",
57
- "value": 0.0832,
58
- "benchmark": "Arena agent steerability",
59
- "version": "2026-09-27",
60
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
61
- "date": "2026-09-27",
62
- "confidence": "adjacent",
63
- "source": "arena",
64
- "note": "arena has it at max; carried to this effort"
65
- },
66
- {
67
- "rung": "claude-fable-5-1#high",
68
- "dim": "frontend",
69
- "value": 1751.4868,
70
- "benchmark": "Arena WebDev rating",
71
- "version": "2026-09-25",
72
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
73
- "date": "2026-09-25",
74
- "confidence": "adjacent",
75
- "source": "arena",
76
- "note": "arena has it at max; carried to this effort"
77
- },
78
- {
79
- "rung": "claude-fable-5-1#low",
80
- "dim": "terminal",
81
- "value": 55.8,
82
- "benchmark": "Terminal-Bench",
83
- "version": "4.0",
84
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
85
- "date": "2026-09-22",
86
- "confidence": "adjacent",
87
- "note": "Terminal-Bench has it at max; carried to this effort"
88
- },
89
- {
90
- "rung": "claude-fable-5-1#low",
91
- "dim": "agentic",
92
- "value": 0.1384,
93
- "benchmark": "Arena agent, net improvement",
94
- "version": "2026-09-27",
95
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
96
- "date": "2026-09-27",
97
- "confidence": "adjacent",
98
- "source": "arena",
99
- "note": "arena has it at max; carried to this effort"
100
- },
101
- {
102
- "rung": "claude-fable-5-1#low",
103
- "dim": "steer",
104
- "value": 0.0832,
105
- "benchmark": "Arena agent steerability",
106
- "version": "2026-09-27",
107
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
108
- "date": "2026-09-27",
109
- "confidence": "adjacent",
110
- "source": "arena",
111
- "note": "arena has it at max; carried to this effort"
112
- },
113
- {
114
- "rung": "claude-fable-5-1#low",
115
- "dim": "frontend",
116
- "value": 1751.4868,
117
- "benchmark": "Arena WebDev rating",
118
- "version": "2026-09-25",
119
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
120
- "date": "2026-09-25",
121
- "confidence": "adjacent",
122
- "source": "arena",
123
- "note": "arena has it at max; carried to this effort"
124
- },
125
- {
126
- "rung": "claude-fable-5-1#max",
127
- "dim": "repo_code",
128
- "value": 67.4,
129
- "benchmark": "DeepSWE",
130
- "version": "1.1",
131
- "url": "https://codingfleet.com/blog/deepswe-v11-leaderboard-2026/",
132
- "date": "2026-09-25",
133
- "confidence": "inferred",
134
- "note": "a 5-trial mean from a secondary source that names no effort"
135
- },
136
- {
137
- "rung": "claude-fable-5-1#max",
138
- "dim": "terminal",
139
- "value": 55.8,
140
- "benchmark": "Terminal-Bench",
141
- "version": "4.0",
142
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
143
- "date": "2026-09-22",
144
- "confidence": "verified"
145
- },
146
- {
147
- "rung": "claude-fable-5-1#max",
148
- "dim": "agentic",
149
- "value": 0.1384,
150
- "benchmark": "Arena agent, net improvement",
151
- "version": "2026-09-27",
152
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
153
- "date": "2026-09-27",
154
- "confidence": "measured",
155
- "source": "arena"
156
- },
157
- {
158
- "rung": "claude-fable-5-1#max",
159
- "dim": "steer",
160
- "value": 0.0832,
161
- "benchmark": "Arena agent steerability",
162
- "version": "2026-09-27",
163
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
164
- "date": "2026-09-27",
165
- "confidence": "measured",
166
- "source": "arena"
167
- },
168
- {
169
- "rung": "claude-fable-5-1#max",
170
- "dim": "frontend",
171
- "value": 1751.4868,
172
- "benchmark": "Arena WebDev rating",
173
- "version": "2026-09-25",
174
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
175
- "date": "2026-09-25",
176
- "confidence": "measured",
177
- "source": "arena"
178
- },
179
- {
180
- "rung": "claude-fable-5-1#medium",
181
- "dim": "terminal",
182
- "value": 55.8,
183
- "benchmark": "Terminal-Bench",
184
- "version": "4.0",
185
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
186
- "date": "2026-09-22",
187
- "confidence": "adjacent",
188
- "note": "Terminal-Bench has it at max; carried to this effort"
189
- },
190
- {
191
- "rung": "claude-fable-5-1#medium",
192
- "dim": "agentic",
193
- "value": 0.1384,
194
- "benchmark": "Arena agent, net improvement",
195
- "version": "2026-09-27",
196
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
197
- "date": "2026-09-27",
198
- "confidence": "adjacent",
199
- "source": "arena",
200
- "note": "arena has it at max; carried to this effort"
201
- },
202
- {
203
- "rung": "claude-fable-5-1#medium",
204
- "dim": "steer",
205
- "value": 0.0832,
206
- "benchmark": "Arena agent steerability",
207
- "version": "2026-09-27",
208
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
209
- "date": "2026-09-27",
210
- "confidence": "adjacent",
211
- "source": "arena",
212
- "note": "arena has it at max; carried to this effort"
213
- },
214
- {
215
- "rung": "claude-fable-5-1#medium",
216
- "dim": "frontend",
217
- "value": 1751.4868,
218
- "benchmark": "Arena WebDev rating",
219
- "version": "2026-09-25",
220
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
221
- "date": "2026-09-25",
222
- "confidence": "adjacent",
223
- "source": "arena",
224
- "note": "arena has it at max; carried to this effort"
225
- },
226
- {
227
- "rung": "claude-fable-5-1#xhigh",
228
- "dim": "terminal",
229
- "value": 55.8,
230
- "benchmark": "Terminal-Bench",
231
- "version": "4.0",
232
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
233
- "date": "2026-09-22",
234
- "confidence": "adjacent",
235
- "note": "Terminal-Bench has it at max; carried to this effort"
236
- },
237
- {
238
- "rung": "claude-fable-5-1#xhigh",
239
- "dim": "agentic",
240
- "value": 0.1384,
241
- "benchmark": "Arena agent, net improvement",
242
- "version": "2026-09-27",
243
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
244
- "date": "2026-09-27",
245
- "confidence": "adjacent",
246
- "source": "arena",
247
- "note": "arena has it at max; carried to this effort"
248
- },
249
- {
250
- "rung": "claude-fable-5-1#xhigh",
251
- "dim": "steer",
252
- "value": 0.0832,
253
- "benchmark": "Arena agent steerability",
254
- "version": "2026-09-27",
255
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
256
- "date": "2026-09-27",
257
- "confidence": "adjacent",
258
- "source": "arena",
259
- "note": "arena has it at max; carried to this effort"
260
- },
261
- {
262
- "rung": "claude-fable-5-1#xhigh",
263
- "dim": "frontend",
264
- "value": 1751.4868,
265
- "benchmark": "Arena WebDev rating",
266
- "version": "2026-09-25",
267
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
268
- "date": "2026-09-25",
269
- "confidence": "adjacent",
270
- "source": "arena",
271
- "note": "arena has it at max; carried to this effort"
272
- },
273
- {
274
- "rung": "claude-haiku-4-5#default",
275
- "dim": "frontend",
276
- "value": 1337.5807,
277
- "benchmark": "WebDev Arena (Epoch AI)",
278
- "version": "2026-09-28",
279
- "url": "https://arena.ai/leaderboard",
280
- "date": "2026-09-28",
281
- "confidence": "calibrated",
282
- "source": "epoch",
283
- "effortAssumed": true,
284
- "fit": {
285
- "source": "epoch",
286
- "field": "webdev",
287
- "a": 0.9655664507608555,
288
- "b": 54.04359475979754,
289
- "r2": 0.9519443341786422,
290
- "n": 61
291
- }
292
- },
293
- {
294
- "rung": "claude-opus-5-5#high",
295
- "dim": "repo_code",
296
- "value": 74.2,
297
- "benchmark": "DeepSWE",
298
- "version": "1.1",
299
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
300
- "date": "2026-09-22",
301
- "confidence": "adjacent",
302
- "note": "DeepSWE has it at max; carried to this effort"
303
- },
304
- {
305
- "rung": "claude-opus-5-5#high",
306
- "dim": "terminal",
307
- "value": 66.4,
308
- "benchmark": "Terminal-Bench",
309
- "version": "4.0",
310
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
311
- "date": "2026-09-22",
312
- "confidence": "adjacent",
313
- "note": "Terminal-Bench has it at xhigh; carried to this effort"
314
- },
315
- {
316
- "rung": "claude-opus-5-5#high",
317
- "dim": "agentic",
318
- "value": 0.1215,
319
- "benchmark": "Arena agent, net improvement",
320
- "version": "2026-09-27",
321
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
322
- "date": "2026-09-27",
323
- "confidence": "measured",
324
- "source": "arena"
325
- },
326
- {
327
- "rung": "claude-opus-5-5#high",
328
- "dim": "steer",
329
- "value": 0.145,
330
- "benchmark": "Arena agent steerability",
331
- "version": "2026-09-27",
332
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
333
- "date": "2026-09-27",
334
- "confidence": "measured",
335
- "source": "arena"
336
- },
337
- {
338
- "rung": "claude-opus-5-5#high",
339
- "dim": "frontend",
340
- "value": 1826.7307,
341
- "benchmark": "Arena WebDev rating",
342
- "version": "2026-09-25",
343
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
344
- "date": "2026-09-25",
345
- "confidence": "adjacent",
346
- "source": "arena",
347
- "note": "arena has it at max; carried to this effort"
348
- },
349
- {
350
- "rung": "claude-opus-5-5#low",
351
- "dim": "repo_code",
352
- "value": 74.2,
353
- "benchmark": "DeepSWE",
354
- "version": "1.1",
355
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
356
- "date": "2026-09-22",
357
- "confidence": "adjacent",
358
- "note": "DeepSWE has it at max; carried to this effort"
359
- },
360
- {
361
- "rung": "claude-opus-5-5#low",
362
- "dim": "terminal",
363
- "value": 66.4,
364
- "benchmark": "Terminal-Bench",
365
- "version": "4.0",
366
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
367
- "date": "2026-09-22",
368
- "confidence": "adjacent",
369
- "note": "Terminal-Bench has it at xhigh; carried to this effort"
370
- },
371
- {
372
- "rung": "claude-opus-5-5#low",
373
- "dim": "agentic",
374
- "value": 0.1215,
375
- "benchmark": "Arena agent, net improvement",
376
- "version": "2026-09-27",
377
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
378
- "date": "2026-09-27",
379
- "confidence": "adjacent",
380
- "source": "arena",
381
- "note": "arena has it at high; carried to this effort"
382
- },
383
- {
384
- "rung": "claude-opus-5-5#low",
385
- "dim": "steer",
386
- "value": 0.145,
387
- "benchmark": "Arena agent steerability",
388
- "version": "2026-09-27",
389
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
390
- "date": "2026-09-27",
391
- "confidence": "adjacent",
392
- "source": "arena",
393
- "note": "arena has it at high; carried to this effort"
394
- },
395
- {
396
- "rung": "claude-opus-5-5#low",
397
- "dim": "frontend",
398
- "value": 1826.7307,
399
- "benchmark": "Arena WebDev rating",
400
- "version": "2026-09-25",
401
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
402
- "date": "2026-09-25",
403
- "confidence": "adjacent",
404
- "source": "arena",
405
- "note": "arena has it at max; carried to this effort"
406
- },
407
- {
408
- "rung": "claude-opus-5-5#max",
409
- "dim": "repo_code",
410
- "value": 74.2,
411
- "benchmark": "DeepSWE",
412
- "version": "1.1",
413
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
414
- "date": "2026-09-22",
415
- "confidence": "verified",
416
- "note": "vendor-run, mean of 5 trials (system card 8.3)"
417
- },
418
- {
419
- "rung": "claude-opus-5-5#max",
420
- "dim": "terminal",
421
- "value": 64.8,
422
- "benchmark": "Terminal-Bench",
423
- "version": "4.0",
424
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
425
- "date": "2026-09-22",
426
- "confidence": "verified"
427
- },
428
- {
429
- "rung": "claude-opus-5-5#max",
430
- "dim": "agentic",
431
- "value": 0.1215,
432
- "benchmark": "Arena agent, net improvement",
433
- "version": "2026-09-27",
434
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
435
- "date": "2026-09-27",
436
- "confidence": "adjacent",
437
- "source": "arena",
438
- "note": "arena has it at high; carried to this effort"
439
- },
440
- {
441
- "rung": "claude-opus-5-5#max",
442
- "dim": "steer",
443
- "value": 0.145,
444
- "benchmark": "Arena agent steerability",
445
- "version": "2026-09-27",
446
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
447
- "date": "2026-09-27",
448
- "confidence": "adjacent",
449
- "source": "arena",
450
- "note": "arena has it at high; carried to this effort"
451
- },
452
- {
453
- "rung": "claude-opus-5-5#max",
454
- "dim": "frontend",
455
- "value": 1826.7307,
456
- "benchmark": "Arena WebDev rating",
457
- "version": "2026-09-25",
458
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
459
- "date": "2026-09-25",
460
- "confidence": "measured",
461
- "source": "arena"
462
- },
463
- {
464
- "rung": "claude-opus-5-5#medium",
465
- "dim": "repo_code",
466
- "value": 74.2,
467
- "benchmark": "DeepSWE",
468
- "version": "1.1",
469
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
470
- "date": "2026-09-22",
471
- "confidence": "adjacent",
472
- "note": "DeepSWE has it at max; carried to this effort"
473
- },
474
- {
475
- "rung": "claude-opus-5-5#medium",
476
- "dim": "terminal",
477
- "value": 66.4,
478
- "benchmark": "Terminal-Bench",
479
- "version": "4.0",
480
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
481
- "date": "2026-09-22",
482
- "confidence": "adjacent",
483
- "note": "Terminal-Bench has it at xhigh; carried to this effort"
484
- },
485
- {
486
- "rung": "claude-opus-5-5#medium",
487
- "dim": "agentic",
488
- "value": 0.1215,
489
- "benchmark": "Arena agent, net improvement",
490
- "version": "2026-09-27",
491
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
492
- "date": "2026-09-27",
493
- "confidence": "adjacent",
494
- "source": "arena",
495
- "note": "arena has it at high; carried to this effort"
496
- },
497
- {
498
- "rung": "claude-opus-5-5#medium",
499
- "dim": "steer",
500
- "value": 0.145,
501
- "benchmark": "Arena agent steerability",
502
- "version": "2026-09-27",
503
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
504
- "date": "2026-09-27",
505
- "confidence": "adjacent",
506
- "source": "arena",
507
- "note": "arena has it at high; carried to this effort"
508
- },
509
- {
510
- "rung": "claude-opus-5-5#medium",
511
- "dim": "frontend",
512
- "value": 1826.7307,
513
- "benchmark": "Arena WebDev rating",
514
- "version": "2026-09-25",
515
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
516
- "date": "2026-09-25",
517
- "confidence": "adjacent",
518
- "source": "arena",
519
- "note": "arena has it at max; carried to this effort"
520
- },
521
- {
522
- "rung": "claude-opus-5-5#xhigh",
523
- "dim": "repo_code",
524
- "value": 74.2,
525
- "benchmark": "DeepSWE",
526
- "version": "1.1",
527
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
528
- "date": "2026-09-22",
529
- "confidence": "adjacent",
530
- "note": "DeepSWE has it at max; carried to this effort"
531
- },
532
- {
533
- "rung": "claude-opus-5-5#xhigh",
534
- "dim": "terminal",
535
- "value": 66.4,
536
- "benchmark": "Terminal-Bench",
537
- "version": "4.0",
538
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
539
- "date": "2026-09-22",
540
- "confidence": "verified",
541
- "note": "Claude Code --bare, 5 trials (system card 8.5)"
542
- },
543
- {
544
- "rung": "claude-opus-5-5#xhigh",
545
- "dim": "agentic",
546
- "value": 0.1215,
547
- "benchmark": "Arena agent, net improvement",
548
- "version": "2026-09-27",
549
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
550
- "date": "2026-09-27",
551
- "confidence": "adjacent",
552
- "source": "arena",
553
- "note": "arena has it at high; carried to this effort"
554
- },
555
- {
556
- "rung": "claude-opus-5-5#xhigh",
557
- "dim": "steer",
558
- "value": 0.145,
559
- "benchmark": "Arena agent steerability",
560
- "version": "2026-09-27",
561
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
562
- "date": "2026-09-27",
563
- "confidence": "adjacent",
564
- "source": "arena",
565
- "note": "arena has it at high; carried to this effort"
566
- },
567
- {
568
- "rung": "claude-opus-5-5#xhigh",
569
- "dim": "frontend",
570
- "value": 1826.7307,
571
- "benchmark": "Arena WebDev rating",
572
- "version": "2026-09-25",
573
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
574
- "date": "2026-09-25",
575
- "confidence": "adjacent",
576
- "source": "arena",
577
- "note": "arena has it at max; carried to this effort"
578
- },
579
- {
580
- "rung": "claude-sonnet-5#high",
581
- "dim": "repo_code",
582
- "value": 54,
583
- "benchmark": "DeepSWE",
584
- "version": "1.1",
585
- "url": "https://deepswe.datacurve.ai/",
586
- "date": "2026-09-22",
587
- "confidence": "adjacent",
588
- "note": "DeepSWE has it at max; carried to this effort"
589
- },
590
- {
591
- "rung": "claude-sonnet-5#high",
592
- "dim": "agentic",
593
- "value": 0.0482,
594
- "benchmark": "Arena agent, net improvement",
595
- "version": "2026-09-27",
596
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
597
- "date": "2026-09-27",
598
- "confidence": "measured",
599
- "source": "arena"
600
- },
601
- {
602
- "rung": "claude-sonnet-5#high",
603
- "dim": "steer",
604
- "value": 0.0744,
605
- "benchmark": "Arena agent steerability",
606
- "version": "2026-09-27",
607
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
608
- "date": "2026-09-27",
609
- "confidence": "measured",
610
- "source": "arena"
611
- },
612
- {
613
- "rung": "claude-sonnet-5#high",
614
- "dim": "frontend",
615
- "value": 1538.9132,
616
- "benchmark": "Arena WebDev rating",
617
- "version": "2026-09-25",
618
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
619
- "date": "2026-09-25",
620
- "confidence": "measured",
621
- "source": "arena"
622
- },
623
- {
624
- "rung": "claude-sonnet-5#low",
625
- "dim": "repo_code",
626
- "value": 54,
627
- "benchmark": "DeepSWE",
628
- "version": "1.1",
629
- "url": "https://deepswe.datacurve.ai/",
630
- "date": "2026-09-22",
631
- "confidence": "adjacent",
632
- "note": "DeepSWE has it at max; carried to this effort"
633
- },
634
- {
635
- "rung": "claude-sonnet-5#low",
636
- "dim": "agentic",
637
- "value": 0.0482,
638
- "benchmark": "Arena agent, net improvement",
639
- "version": "2026-09-27",
640
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
641
- "date": "2026-09-27",
642
- "confidence": "adjacent",
643
- "source": "arena",
644
- "note": "arena has it at high; carried to this effort"
645
- },
646
- {
647
- "rung": "claude-sonnet-5#low",
648
- "dim": "steer",
649
- "value": 0.0744,
650
- "benchmark": "Arena agent steerability",
651
- "version": "2026-09-27",
652
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
653
- "date": "2026-09-27",
654
- "confidence": "adjacent",
655
- "source": "arena",
656
- "note": "arena has it at high; carried to this effort"
657
- },
658
- {
659
- "rung": "claude-sonnet-5#low",
660
- "dim": "frontend",
661
- "value": 1538.9132,
662
- "benchmark": "Arena WebDev rating",
663
- "version": "2026-09-25",
664
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
665
- "date": "2026-09-25",
666
- "confidence": "adjacent",
667
- "source": "arena",
668
- "note": "arena has it at high; carried to this effort"
669
- },
670
- {
671
- "rung": "claude-sonnet-5#max",
672
- "dim": "repo_code",
673
- "value": 54,
674
- "benchmark": "DeepSWE",
675
- "version": "1.1",
676
- "url": "https://deepswe.datacurve.ai/",
677
- "date": "2026-09-22",
678
- "confidence": "verified"
679
- },
680
- {
681
- "rung": "claude-sonnet-5#max",
682
- "dim": "agentic",
683
- "value": 0.0482,
684
- "benchmark": "Arena agent, net improvement",
685
- "version": "2026-09-27",
686
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
687
- "date": "2026-09-27",
688
- "confidence": "adjacent",
689
- "source": "arena",
690
- "note": "arena has it at high; carried to this effort"
691
- },
692
- {
693
- "rung": "claude-sonnet-5#max",
694
- "dim": "steer",
695
- "value": 0.0744,
696
- "benchmark": "Arena agent steerability",
697
- "version": "2026-09-27",
698
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
699
- "date": "2026-09-27",
700
- "confidence": "adjacent",
701
- "source": "arena",
702
- "note": "arena has it at high; carried to this effort"
703
- },
704
- {
705
- "rung": "claude-sonnet-5#max",
706
- "dim": "frontend",
707
- "value": 1538.9132,
708
- "benchmark": "Arena WebDev rating",
709
- "version": "2026-09-25",
710
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
711
- "date": "2026-09-25",
712
- "confidence": "adjacent",
713
- "source": "arena",
714
- "note": "arena has it at high; carried to this effort"
715
- },
716
- {
717
- "rung": "claude-sonnet-5#medium",
718
- "dim": "repo_code",
719
- "value": 54,
720
- "benchmark": "DeepSWE",
721
- "version": "1.1",
722
- "url": "https://deepswe.datacurve.ai/",
723
- "date": "2026-09-22",
724
- "confidence": "adjacent",
725
- "note": "DeepSWE has it at max; carried to this effort"
726
- },
727
- {
728
- "rung": "claude-sonnet-5#medium",
729
- "dim": "agentic",
730
- "value": 0.0482,
731
- "benchmark": "Arena agent, net improvement",
732
- "version": "2026-09-27",
733
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
734
- "date": "2026-09-27",
735
- "confidence": "adjacent",
736
- "source": "arena",
737
- "note": "arena has it at high; carried to this effort"
738
- },
739
- {
740
- "rung": "claude-sonnet-5#medium",
741
- "dim": "steer",
742
- "value": 0.0744,
743
- "benchmark": "Arena agent steerability",
744
- "version": "2026-09-27",
745
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
746
- "date": "2026-09-27",
747
- "confidence": "adjacent",
748
- "source": "arena",
749
- "note": "arena has it at high; carried to this effort"
750
- },
751
- {
752
- "rung": "claude-sonnet-5#medium",
753
- "dim": "frontend",
754
- "value": 1538.9132,
755
- "benchmark": "Arena WebDev rating",
756
- "version": "2026-09-25",
757
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
758
- "date": "2026-09-25",
759
- "confidence": "adjacent",
760
- "source": "arena",
761
- "note": "arena has it at high; carried to this effort"
762
- },
763
- {
764
- "rung": "claude-sonnet-5#xhigh",
765
- "dim": "repo_code",
766
- "value": 54,
767
- "benchmark": "DeepSWE",
768
- "version": "1.1",
769
- "url": "https://deepswe.datacurve.ai/",
770
- "date": "2026-09-22",
771
- "confidence": "adjacent",
772
- "note": "DeepSWE has it at max; carried to this effort"
773
- },
774
- {
775
- "rung": "claude-sonnet-5#xhigh",
776
- "dim": "agentic",
777
- "value": 0.0482,
778
- "benchmark": "Arena agent, net improvement",
779
- "version": "2026-09-27",
780
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
781
- "date": "2026-09-27",
782
- "confidence": "adjacent",
783
- "source": "arena",
784
- "note": "arena has it at high; carried to this effort"
785
- },
786
- {
787
- "rung": "claude-sonnet-5#xhigh",
788
- "dim": "steer",
789
- "value": 0.0744,
790
- "benchmark": "Arena agent steerability",
791
- "version": "2026-09-27",
792
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
793
- "date": "2026-09-27",
794
- "confidence": "adjacent",
795
- "source": "arena",
796
- "note": "arena has it at high; carried to this effort"
797
- },
798
- {
799
- "rung": "claude-sonnet-5#xhigh",
800
- "dim": "frontend",
801
- "value": 1538.9132,
802
- "benchmark": "Arena WebDev rating",
803
- "version": "2026-09-25",
804
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
805
- "date": "2026-09-25",
806
- "confidence": "adjacent",
807
- "source": "arena",
808
- "note": "arena has it at high; carried to this effort"
809
- },
810
- {
811
- "rung": "gpt-5.6-luna#high",
812
- "dim": "repo_code",
813
- "value": 67,
814
- "benchmark": "DeepSWE",
815
- "version": "1.1",
816
- "url": "https://deepswe.datacurve.ai/",
817
- "date": "2026-09-22",
818
- "confidence": "adjacent",
819
- "note": "DeepSWE has it at max; carried to this effort"
820
- },
821
- {
822
- "rung": "gpt-5.6-luna#high",
823
- "dim": "honesty",
824
- "value": 21.8,
825
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
826
- "version": "GPT-6 system card appendix 11.6.4.2",
827
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
828
- "date": "2026-09-22",
829
- "confidence": "adjacent",
830
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
831
- },
832
- {
833
- "rung": "gpt-5.6-luna#high",
834
- "dim": "agentic",
835
- "value": -0.0075,
836
- "benchmark": "Arena agent, net improvement",
837
- "version": "2026-09-27",
838
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
839
- "date": "2026-09-27",
840
- "confidence": "adjacent",
841
- "source": "arena",
842
- "note": "arena has it at xhigh; carried to this effort"
843
- },
844
- {
845
- "rung": "gpt-5.6-luna#high",
846
- "dim": "steer",
847
- "value": 0.0185,
848
- "benchmark": "Arena agent steerability",
849
- "version": "2026-09-27",
850
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
851
- "date": "2026-09-27",
852
- "confidence": "adjacent",
853
- "source": "arena",
854
- "note": "arena has it at xhigh; carried to this effort"
855
- },
856
- {
857
- "rung": "gpt-5.6-luna#high",
858
- "dim": "frontend",
859
- "value": 1519.8875,
860
- "benchmark": "Arena WebDev rating",
861
- "version": "2026-09-25",
862
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
863
- "date": "2026-09-25",
864
- "confidence": "adjacent",
865
- "source": "arena",
866
- "note": "arena has it at xhigh; carried to this effort"
867
- },
868
- {
869
- "rung": "gpt-5.6-luna#low",
870
- "dim": "repo_code",
871
- "value": 67,
872
- "benchmark": "DeepSWE",
873
- "version": "1.1",
874
- "url": "https://deepswe.datacurve.ai/",
875
- "date": "2026-09-22",
876
- "confidence": "adjacent",
877
- "note": "DeepSWE has it at max; carried to this effort"
878
- },
879
- {
880
- "rung": "gpt-5.6-luna#low",
881
- "dim": "honesty",
882
- "value": 21.8,
883
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
884
- "version": "GPT-6 system card appendix 11.6.4.2",
885
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
886
- "date": "2026-09-22",
887
- "confidence": "adjacent",
888
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
889
- },
890
- {
891
- "rung": "gpt-5.6-luna#low",
892
- "dim": "agentic",
893
- "value": -0.0075,
894
- "benchmark": "Arena agent, net improvement",
895
- "version": "2026-09-27",
896
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
897
- "date": "2026-09-27",
898
- "confidence": "adjacent",
899
- "source": "arena",
900
- "note": "arena has it at xhigh; carried to this effort"
901
- },
902
- {
903
- "rung": "gpt-5.6-luna#low",
904
- "dim": "steer",
905
- "value": 0.0185,
906
- "benchmark": "Arena agent steerability",
907
- "version": "2026-09-27",
908
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
909
- "date": "2026-09-27",
910
- "confidence": "adjacent",
911
- "source": "arena",
912
- "note": "arena has it at xhigh; carried to this effort"
913
- },
914
- {
915
- "rung": "gpt-5.6-luna#low",
916
- "dim": "frontend",
917
- "value": 1519.8875,
918
- "benchmark": "Arena WebDev rating",
919
- "version": "2026-09-25",
920
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
921
- "date": "2026-09-25",
922
- "confidence": "adjacent",
923
- "source": "arena",
924
- "note": "arena has it at xhigh; carried to this effort"
925
- },
926
- {
927
- "rung": "gpt-5.6-luna#max",
928
- "dim": "repo_code",
929
- "value": 67,
930
- "benchmark": "DeepSWE",
931
- "version": "1.1",
932
- "url": "https://deepswe.datacurve.ai/",
933
- "date": "2026-09-22",
934
- "confidence": "verified"
935
- },
936
- {
937
- "rung": "gpt-5.6-luna#max",
938
- "dim": "honesty",
939
- "value": 21.8,
940
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
941
- "version": "GPT-6 system card appendix 11.6.4.2",
942
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
943
- "date": "2026-09-22",
944
- "confidence": "verified"
945
- },
946
- {
947
- "rung": "gpt-5.6-luna#max",
948
- "dim": "agentic",
949
- "value": -0.0075,
950
- "benchmark": "Arena agent, net improvement",
951
- "version": "2026-09-27",
952
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
953
- "date": "2026-09-27",
954
- "confidence": "adjacent",
955
- "source": "arena",
956
- "note": "arena has it at xhigh; carried to this effort"
957
- },
958
- {
959
- "rung": "gpt-5.6-luna#max",
960
- "dim": "steer",
961
- "value": 0.0185,
962
- "benchmark": "Arena agent steerability",
963
- "version": "2026-09-27",
964
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
965
- "date": "2026-09-27",
966
- "confidence": "adjacent",
967
- "source": "arena",
968
- "note": "arena has it at xhigh; carried to this effort"
969
- },
970
- {
971
- "rung": "gpt-5.6-luna#max",
972
- "dim": "frontend",
973
- "value": 1519.8875,
974
- "benchmark": "Arena WebDev rating",
975
- "version": "2026-09-25",
976
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
977
- "date": "2026-09-25",
978
- "confidence": "adjacent",
979
- "source": "arena",
980
- "note": "arena has it at xhigh; carried to this effort"
981
- },
982
- {
983
- "rung": "gpt-5.6-luna#medium",
984
- "dim": "repo_code",
985
- "value": 67,
986
- "benchmark": "DeepSWE",
987
- "version": "1.1",
988
- "url": "https://deepswe.datacurve.ai/",
989
- "date": "2026-09-22",
990
- "confidence": "adjacent",
991
- "note": "DeepSWE has it at max; carried to this effort"
992
- },
993
- {
994
- "rung": "gpt-5.6-luna#medium",
995
- "dim": "honesty",
996
- "value": 21.8,
997
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
998
- "version": "GPT-6 system card appendix 11.6.4.2",
999
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
1000
- "date": "2026-09-22",
1001
- "confidence": "adjacent",
1002
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
1003
- },
1004
- {
1005
- "rung": "gpt-5.6-luna#medium",
1006
- "dim": "agentic",
1007
- "value": -0.0075,
1008
- "benchmark": "Arena agent, net improvement",
1009
- "version": "2026-09-27",
1010
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1011
- "date": "2026-09-27",
1012
- "confidence": "adjacent",
1013
- "source": "arena",
1014
- "note": "arena has it at xhigh; carried to this effort"
1015
- },
1016
- {
1017
- "rung": "gpt-5.6-luna#medium",
1018
- "dim": "steer",
1019
- "value": 0.0185,
1020
- "benchmark": "Arena agent steerability",
1021
- "version": "2026-09-27",
1022
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1023
- "date": "2026-09-27",
1024
- "confidence": "adjacent",
1025
- "source": "arena",
1026
- "note": "arena has it at xhigh; carried to this effort"
1027
- },
1028
- {
1029
- "rung": "gpt-5.6-luna#medium",
1030
- "dim": "frontend",
1031
- "value": 1519.8875,
1032
- "benchmark": "Arena WebDev rating",
1033
- "version": "2026-09-25",
1034
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1035
- "date": "2026-09-25",
1036
- "confidence": "adjacent",
1037
- "source": "arena",
1038
- "note": "arena has it at xhigh; carried to this effort"
1039
- },
1040
- {
1041
- "rung": "gpt-5.6-luna#none",
1042
- "dim": "repo_code",
1043
- "value": 67,
1044
- "benchmark": "DeepSWE",
1045
- "version": "1.1",
1046
- "url": "https://deepswe.datacurve.ai/",
1047
- "date": "2026-09-22",
1048
- "confidence": "adjacent",
1049
- "note": "DeepSWE has it at max; carried to this effort"
1050
- },
1051
- {
1052
- "rung": "gpt-5.6-luna#none",
1053
- "dim": "honesty",
1054
- "value": 21.8,
1055
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
1056
- "version": "GPT-6 system card appendix 11.6.4.2",
1057
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
1058
- "date": "2026-09-22",
1059
- "confidence": "adjacent",
1060
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
1061
- },
1062
- {
1063
- "rung": "gpt-5.6-luna#none",
1064
- "dim": "agentic",
1065
- "value": -0.0075,
1066
- "benchmark": "Arena agent, net improvement",
1067
- "version": "2026-09-27",
1068
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1069
- "date": "2026-09-27",
1070
- "confidence": "adjacent",
1071
- "source": "arena",
1072
- "note": "arena has it at xhigh; carried to this effort"
1073
- },
1074
- {
1075
- "rung": "gpt-5.6-luna#none",
1076
- "dim": "steer",
1077
- "value": 0.0185,
1078
- "benchmark": "Arena agent steerability",
1079
- "version": "2026-09-27",
1080
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1081
- "date": "2026-09-27",
1082
- "confidence": "adjacent",
1083
- "source": "arena",
1084
- "note": "arena has it at xhigh; carried to this effort"
1085
- },
1086
- {
1087
- "rung": "gpt-5.6-luna#none",
1088
- "dim": "frontend",
1089
- "value": 1519.8875,
1090
- "benchmark": "Arena WebDev rating",
1091
- "version": "2026-09-25",
1092
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1093
- "date": "2026-09-25",
1094
- "confidence": "adjacent",
1095
- "source": "arena",
1096
- "note": "arena has it at xhigh; carried to this effort"
1097
- },
1098
- {
1099
- "rung": "gpt-5.6-luna#xhigh",
1100
- "dim": "repo_code",
1101
- "value": 67,
1102
- "benchmark": "DeepSWE",
1103
- "version": "1.1",
1104
- "url": "https://deepswe.datacurve.ai/",
1105
- "date": "2026-09-22",
1106
- "confidence": "adjacent",
1107
- "note": "DeepSWE has it at max; carried to this effort"
1108
- },
1109
- {
1110
- "rung": "gpt-5.6-luna#xhigh",
1111
- "dim": "honesty",
1112
- "value": 21.8,
1113
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
1114
- "version": "GPT-6 system card appendix 11.6.4.2",
1115
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
1116
- "date": "2026-09-22",
1117
- "confidence": "adjacent",
1118
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
1119
- },
1120
- {
1121
- "rung": "gpt-5.6-luna#xhigh",
1122
- "dim": "agentic",
1123
- "value": -0.0075,
1124
- "benchmark": "Arena agent, net improvement",
1125
- "version": "2026-09-27",
1126
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1127
- "date": "2026-09-27",
1128
- "confidence": "measured",
1129
- "source": "arena"
1130
- },
1131
- {
1132
- "rung": "gpt-5.6-luna#xhigh",
1133
- "dim": "steer",
1134
- "value": 0.0185,
1135
- "benchmark": "Arena agent steerability",
1136
- "version": "2026-09-27",
1137
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1138
- "date": "2026-09-27",
1139
- "confidence": "measured",
1140
- "source": "arena"
1141
- },
1142
- {
1143
- "rung": "gpt-5.6-luna#xhigh",
1144
- "dim": "frontend",
1145
- "value": 1519.8875,
1146
- "benchmark": "Arena WebDev rating",
1147
- "version": "2026-09-25",
1148
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1149
- "date": "2026-09-25",
1150
- "confidence": "measured",
1151
- "source": "arena"
1152
- },
1153
- {
1154
- "rung": "gpt-5.6-sol#high",
1155
- "dim": "repo_code",
1156
- "value": 73,
1157
- "benchmark": "DeepSWE",
1158
- "version": "1.1",
1159
- "url": "https://deepswe.datacurve.ai/",
1160
- "date": "2026-09-22",
1161
- "confidence": "adjacent",
1162
- "note": "DeepSWE has it at max; carried to this effort"
1163
- },
1164
- {
1165
- "rung": "gpt-5.6-sol#high",
1166
- "dim": "terminal",
1167
- "value": 37.3,
1168
- "benchmark": "Terminal-Bench",
1169
- "version": "4.0",
1170
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
1171
- "date": "2026-09-22",
1172
- "confidence": "adjacent",
1173
- "note": "Terminal-Bench has it at max; carried to this effort"
1174
- },
1175
- {
1176
- "rung": "gpt-5.6-sol#high",
1177
- "dim": "honesty",
1178
- "value": 22.5,
1179
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
1180
- "version": "GPT-6 system card appendix 11.6.4.2",
1181
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
1182
- "date": "2026-09-22",
1183
- "confidence": "adjacent",
1184
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
1185
- },
1186
- {
1187
- "rung": "gpt-5.6-sol#high",
1188
- "dim": "agentic",
1189
- "value": 0.0621,
1190
- "benchmark": "Arena agent, net improvement",
1191
- "version": "2026-09-27",
1192
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1193
- "date": "2026-09-27",
1194
- "confidence": "adjacent",
1195
- "source": "arena",
1196
- "note": "arena has it at xhigh; carried to this effort"
1197
- },
1198
- {
1199
- "rung": "gpt-5.6-sol#high",
1200
- "dim": "steer",
1201
- "value": 0.0656,
1202
- "benchmark": "Arena agent steerability",
1203
- "version": "2026-09-27",
1204
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1205
- "date": "2026-09-27",
1206
- "confidence": "adjacent",
1207
- "source": "arena",
1208
- "note": "arena has it at xhigh; carried to this effort"
1209
- },
1210
- {
1211
- "rung": "gpt-5.6-sol#high",
1212
- "dim": "frontend",
1213
- "value": 1617.3525,
1214
- "benchmark": "Arena WebDev rating",
1215
- "version": "2026-09-25",
1216
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1217
- "date": "2026-09-25",
1218
- "confidence": "adjacent",
1219
- "source": "arena",
1220
- "note": "arena has it at xhigh; carried to this effort"
1221
- },
1222
- {
1223
- "rung": "gpt-5.6-sol#low",
1224
- "dim": "repo_code",
1225
- "value": 73,
1226
- "benchmark": "DeepSWE",
1227
- "version": "1.1",
1228
- "url": "https://deepswe.datacurve.ai/",
1229
- "date": "2026-09-22",
1230
- "confidence": "adjacent",
1231
- "note": "DeepSWE has it at max; carried to this effort"
1232
- },
1233
- {
1234
- "rung": "gpt-5.6-sol#low",
1235
- "dim": "terminal",
1236
- "value": 37.3,
1237
- "benchmark": "Terminal-Bench",
1238
- "version": "4.0",
1239
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
1240
- "date": "2026-09-22",
1241
- "confidence": "adjacent",
1242
- "note": "Terminal-Bench has it at max; carried to this effort"
1243
- },
1244
- {
1245
- "rung": "gpt-5.6-sol#low",
1246
- "dim": "honesty",
1247
- "value": 22.5,
1248
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
1249
- "version": "GPT-6 system card appendix 11.6.4.2",
1250
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
1251
- "date": "2026-09-22",
1252
- "confidence": "adjacent",
1253
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
1254
- },
1255
- {
1256
- "rung": "gpt-5.6-sol#low",
1257
- "dim": "agentic",
1258
- "value": 0.0621,
1259
- "benchmark": "Arena agent, net improvement",
1260
- "version": "2026-09-27",
1261
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1262
- "date": "2026-09-27",
1263
- "confidence": "adjacent",
1264
- "source": "arena",
1265
- "note": "arena has it at xhigh; carried to this effort"
1266
- },
1267
- {
1268
- "rung": "gpt-5.6-sol#low",
1269
- "dim": "steer",
1270
- "value": 0.0656,
1271
- "benchmark": "Arena agent steerability",
1272
- "version": "2026-09-27",
1273
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1274
- "date": "2026-09-27",
1275
- "confidence": "adjacent",
1276
- "source": "arena",
1277
- "note": "arena has it at xhigh; carried to this effort"
1278
- },
1279
- {
1280
- "rung": "gpt-5.6-sol#low",
1281
- "dim": "frontend",
1282
- "value": 1617.3525,
1283
- "benchmark": "Arena WebDev rating",
1284
- "version": "2026-09-25",
1285
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1286
- "date": "2026-09-25",
1287
- "confidence": "adjacent",
1288
- "source": "arena",
1289
- "note": "arena has it at xhigh; carried to this effort"
1290
- },
1291
- {
1292
- "rung": "gpt-5.6-sol#max",
1293
- "dim": "repo_code",
1294
- "value": 73,
1295
- "benchmark": "DeepSWE",
1296
- "version": "1.1",
1297
- "url": "https://deepswe.datacurve.ai/",
1298
- "date": "2026-09-22",
1299
- "confidence": "verified"
1300
- },
1301
- {
1302
- "rung": "gpt-5.6-sol#max",
1303
- "dim": "terminal",
1304
- "value": 37.3,
1305
- "benchmark": "Terminal-Bench",
1306
- "version": "4.0",
1307
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
1308
- "date": "2026-09-22",
1309
- "confidence": "secondary"
1310
- },
1311
- {
1312
- "rung": "gpt-5.6-sol#max",
1313
- "dim": "honesty",
1314
- "value": 22.5,
1315
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
1316
- "version": "GPT-6 system card appendix 11.6.4.2",
1317
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
1318
- "date": "2026-09-22",
1319
- "confidence": "verified"
1320
- },
1321
- {
1322
- "rung": "gpt-5.6-sol#max",
1323
- "dim": "agentic",
1324
- "value": 0.0621,
1325
- "benchmark": "Arena agent, net improvement",
1326
- "version": "2026-09-27",
1327
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1328
- "date": "2026-09-27",
1329
- "confidence": "adjacent",
1330
- "source": "arena",
1331
- "note": "arena has it at xhigh; carried to this effort"
1332
- },
1333
- {
1334
- "rung": "gpt-5.6-sol#max",
1335
- "dim": "steer",
1336
- "value": 0.0656,
1337
- "benchmark": "Arena agent steerability",
1338
- "version": "2026-09-27",
1339
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1340
- "date": "2026-09-27",
1341
- "confidence": "adjacent",
1342
- "source": "arena",
1343
- "note": "arena has it at xhigh; carried to this effort"
1344
- },
1345
- {
1346
- "rung": "gpt-5.6-sol#max",
1347
- "dim": "frontend",
1348
- "value": 1617.3525,
1349
- "benchmark": "Arena WebDev rating",
1350
- "version": "2026-09-25",
1351
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1352
- "date": "2026-09-25",
1353
- "confidence": "adjacent",
1354
- "source": "arena",
1355
- "note": "arena has it at xhigh; carried to this effort"
1356
- },
1357
- {
1358
- "rung": "gpt-5.6-sol#medium",
1359
- "dim": "repo_code",
1360
- "value": 73,
1361
- "benchmark": "DeepSWE",
1362
- "version": "1.1",
1363
- "url": "https://deepswe.datacurve.ai/",
1364
- "date": "2026-09-22",
1365
- "confidence": "adjacent",
1366
- "note": "DeepSWE has it at max; carried to this effort"
1367
- },
1368
- {
1369
- "rung": "gpt-5.6-sol#medium",
1370
- "dim": "terminal",
1371
- "value": 37.3,
1372
- "benchmark": "Terminal-Bench",
1373
- "version": "4.0",
1374
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
1375
- "date": "2026-09-22",
1376
- "confidence": "adjacent",
1377
- "note": "Terminal-Bench has it at max; carried to this effort"
1378
- },
1379
- {
1380
- "rung": "gpt-5.6-sol#medium",
1381
- "dim": "honesty",
1382
- "value": 22.5,
1383
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
1384
- "version": "GPT-6 system card appendix 11.6.4.2",
1385
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
1386
- "date": "2026-09-22",
1387
- "confidence": "adjacent",
1388
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
1389
- },
1390
- {
1391
- "rung": "gpt-5.6-sol#medium",
1392
- "dim": "agentic",
1393
- "value": 0.0621,
1394
- "benchmark": "Arena agent, net improvement",
1395
- "version": "2026-09-27",
1396
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1397
- "date": "2026-09-27",
1398
- "confidence": "adjacent",
1399
- "source": "arena",
1400
- "note": "arena has it at xhigh; carried to this effort"
1401
- },
1402
- {
1403
- "rung": "gpt-5.6-sol#medium",
1404
- "dim": "steer",
1405
- "value": 0.0656,
1406
- "benchmark": "Arena agent steerability",
1407
- "version": "2026-09-27",
1408
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1409
- "date": "2026-09-27",
1410
- "confidence": "adjacent",
1411
- "source": "arena",
1412
- "note": "arena has it at xhigh; carried to this effort"
1413
- },
1414
- {
1415
- "rung": "gpt-5.6-sol#medium",
1416
- "dim": "frontend",
1417
- "value": 1617.3525,
1418
- "benchmark": "Arena WebDev rating",
1419
- "version": "2026-09-25",
1420
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1421
- "date": "2026-09-25",
1422
- "confidence": "adjacent",
1423
- "source": "arena",
1424
- "note": "arena has it at xhigh; carried to this effort"
1425
- },
1426
- {
1427
- "rung": "gpt-5.6-sol#none",
1428
- "dim": "repo_code",
1429
- "value": 73,
1430
- "benchmark": "DeepSWE",
1431
- "version": "1.1",
1432
- "url": "https://deepswe.datacurve.ai/",
1433
- "date": "2026-09-22",
1434
- "confidence": "adjacent",
1435
- "note": "DeepSWE has it at max; carried to this effort"
1436
- },
1437
- {
1438
- "rung": "gpt-5.6-sol#none",
1439
- "dim": "terminal",
1440
- "value": 37.3,
1441
- "benchmark": "Terminal-Bench",
1442
- "version": "4.0",
1443
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
1444
- "date": "2026-09-22",
1445
- "confidence": "adjacent",
1446
- "note": "Terminal-Bench has it at max; carried to this effort"
1447
- },
1448
- {
1449
- "rung": "gpt-5.6-sol#none",
1450
- "dim": "honesty",
1451
- "value": 22.5,
1452
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
1453
- "version": "GPT-6 system card appendix 11.6.4.2",
1454
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
1455
- "date": "2026-09-22",
1456
- "confidence": "adjacent",
1457
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
1458
- },
1459
- {
1460
- "rung": "gpt-5.6-sol#none",
1461
- "dim": "agentic",
1462
- "value": 0.0621,
1463
- "benchmark": "Arena agent, net improvement",
1464
- "version": "2026-09-27",
1465
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1466
- "date": "2026-09-27",
1467
- "confidence": "adjacent",
1468
- "source": "arena",
1469
- "note": "arena has it at xhigh; carried to this effort"
1470
- },
1471
- {
1472
- "rung": "gpt-5.6-sol#none",
1473
- "dim": "steer",
1474
- "value": 0.0656,
1475
- "benchmark": "Arena agent steerability",
1476
- "version": "2026-09-27",
1477
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1478
- "date": "2026-09-27",
1479
- "confidence": "adjacent",
1480
- "source": "arena",
1481
- "note": "arena has it at xhigh; carried to this effort"
1482
- },
1483
- {
1484
- "rung": "gpt-5.6-sol#none",
1485
- "dim": "frontend",
1486
- "value": 1617.3525,
1487
- "benchmark": "Arena WebDev rating",
1488
- "version": "2026-09-25",
1489
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1490
- "date": "2026-09-25",
1491
- "confidence": "adjacent",
1492
- "source": "arena",
1493
- "note": "arena has it at xhigh; carried to this effort"
1494
- },
1495
- {
1496
- "rung": "gpt-5.6-sol#ultra",
1497
- "dim": "repo_code",
1498
- "value": 73,
1499
- "benchmark": "DeepSWE",
1500
- "version": "1.1",
1501
- "url": "https://deepswe.datacurve.ai/",
1502
- "date": "2026-09-22",
1503
- "confidence": "adjacent",
1504
- "note": "DeepSWE has it at max; carried to this effort"
1505
- },
1506
- {
1507
- "rung": "gpt-5.6-sol#ultra",
1508
- "dim": "terminal",
1509
- "value": 37.3,
1510
- "benchmark": "Terminal-Bench",
1511
- "version": "4.0",
1512
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
1513
- "date": "2026-09-22",
1514
- "confidence": "adjacent",
1515
- "note": "Terminal-Bench has it at max; carried to this effort"
1516
- },
1517
- {
1518
- "rung": "gpt-5.6-sol#ultra",
1519
- "dim": "honesty",
1520
- "value": 22.5,
1521
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
1522
- "version": "GPT-6 system card appendix 11.6.4.2",
1523
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
1524
- "date": "2026-09-22",
1525
- "confidence": "adjacent",
1526
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
1527
- },
1528
- {
1529
- "rung": "gpt-5.6-sol#ultra",
1530
- "dim": "agentic",
1531
- "value": 0.0621,
1532
- "benchmark": "Arena agent, net improvement",
1533
- "version": "2026-09-27",
1534
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1535
- "date": "2026-09-27",
1536
- "confidence": "adjacent",
1537
- "source": "arena",
1538
- "note": "arena has it at xhigh; carried to this effort"
1539
- },
1540
- {
1541
- "rung": "gpt-5.6-sol#ultra",
1542
- "dim": "steer",
1543
- "value": 0.0656,
1544
- "benchmark": "Arena agent steerability",
1545
- "version": "2026-09-27",
1546
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1547
- "date": "2026-09-27",
1548
- "confidence": "adjacent",
1549
- "source": "arena",
1550
- "note": "arena has it at xhigh; carried to this effort"
1551
- },
1552
- {
1553
- "rung": "gpt-5.6-sol#ultra",
1554
- "dim": "frontend",
1555
- "value": 1617.3525,
1556
- "benchmark": "Arena WebDev rating",
1557
- "version": "2026-09-25",
1558
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1559
- "date": "2026-09-25",
1560
- "confidence": "adjacent",
1561
- "source": "arena",
1562
- "note": "arena has it at xhigh; carried to this effort"
1563
- },
1564
- {
1565
- "rung": "gpt-5.6-sol#xhigh",
1566
- "dim": "repo_code",
1567
- "value": 73,
1568
- "benchmark": "DeepSWE",
1569
- "version": "1.1",
1570
- "url": "https://deepswe.datacurve.ai/",
1571
- "date": "2026-09-22",
1572
- "confidence": "adjacent",
1573
- "note": "DeepSWE has it at max; carried to this effort"
1574
- },
1575
- {
1576
- "rung": "gpt-5.6-sol#xhigh",
1577
- "dim": "terminal",
1578
- "value": 37.3,
1579
- "benchmark": "Terminal-Bench",
1580
- "version": "4.0",
1581
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
1582
- "date": "2026-09-22",
1583
- "confidence": "adjacent",
1584
- "note": "Terminal-Bench has it at max; carried to this effort"
1585
- },
1586
- {
1587
- "rung": "gpt-5.6-sol#xhigh",
1588
- "dim": "honesty",
1589
- "value": 22.5,
1590
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
1591
- "version": "GPT-6 system card appendix 11.6.4.2",
1592
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
1593
- "date": "2026-09-22",
1594
- "confidence": "adjacent",
1595
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
1596
- },
1597
- {
1598
- "rung": "gpt-5.6-sol#xhigh",
1599
- "dim": "agentic",
1600
- "value": 0.0621,
1601
- "benchmark": "Arena agent, net improvement",
1602
- "version": "2026-09-27",
1603
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1604
- "date": "2026-09-27",
1605
- "confidence": "measured",
1606
- "source": "arena"
1607
- },
1608
- {
1609
- "rung": "gpt-5.6-sol#xhigh",
1610
- "dim": "steer",
1611
- "value": 0.0656,
1612
- "benchmark": "Arena agent steerability",
1613
- "version": "2026-09-27",
1614
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1615
- "date": "2026-09-27",
1616
- "confidence": "measured",
1617
- "source": "arena"
1618
- },
1619
- {
1620
- "rung": "gpt-5.6-sol#xhigh",
1621
- "dim": "frontend",
1622
- "value": 1617.3525,
1623
- "benchmark": "Arena WebDev rating",
1624
- "version": "2026-09-25",
1625
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1626
- "date": "2026-09-25",
1627
- "confidence": "measured",
1628
- "source": "arena"
1629
- },
1630
- {
1631
- "rung": "gpt-5.6-terra#high",
1632
- "dim": "agentic",
1633
- "value": 0.0071,
1634
- "benchmark": "Arena agent, net improvement",
1635
- "version": "2026-09-27",
1636
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1637
- "date": "2026-09-27",
1638
- "confidence": "adjacent",
1639
- "source": "arena",
1640
- "note": "arena has it at xhigh; carried to this effort"
1641
- },
1642
- {
1643
- "rung": "gpt-5.6-terra#high",
1644
- "dim": "steer",
1645
- "value": 0.0423,
1646
- "benchmark": "Arena agent steerability",
1647
- "version": "2026-09-27",
1648
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1649
- "date": "2026-09-27",
1650
- "confidence": "adjacent",
1651
- "source": "arena",
1652
- "note": "arena has it at xhigh; carried to this effort"
1653
- },
1654
- {
1655
- "rung": "gpt-5.6-terra#high",
1656
- "dim": "frontend",
1657
- "value": 1520.5493,
1658
- "benchmark": "Arena WebDev rating",
1659
- "version": "2026-09-25",
1660
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1661
- "date": "2026-09-25",
1662
- "confidence": "adjacent",
1663
- "source": "arena",
1664
- "note": "arena has it at xhigh; carried to this effort"
1665
- },
1666
- {
1667
- "rung": "gpt-5.6-terra#low",
1668
- "dim": "agentic",
1669
- "value": 0.0071,
1670
- "benchmark": "Arena agent, net improvement",
1671
- "version": "2026-09-27",
1672
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1673
- "date": "2026-09-27",
1674
- "confidence": "adjacent",
1675
- "source": "arena",
1676
- "note": "arena has it at xhigh; carried to this effort"
1677
- },
1678
- {
1679
- "rung": "gpt-5.6-terra#low",
1680
- "dim": "steer",
1681
- "value": 0.0423,
1682
- "benchmark": "Arena agent steerability",
1683
- "version": "2026-09-27",
1684
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1685
- "date": "2026-09-27",
1686
- "confidence": "adjacent",
1687
- "source": "arena",
1688
- "note": "arena has it at xhigh; carried to this effort"
1689
- },
1690
- {
1691
- "rung": "gpt-5.6-terra#low",
1692
- "dim": "frontend",
1693
- "value": 1520.5493,
1694
- "benchmark": "Arena WebDev rating",
1695
- "version": "2026-09-25",
1696
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1697
- "date": "2026-09-25",
1698
- "confidence": "adjacent",
1699
- "source": "arena",
1700
- "note": "arena has it at xhigh; carried to this effort"
1701
- },
1702
- {
1703
- "rung": "gpt-5.6-terra#max",
1704
- "dim": "agentic",
1705
- "value": 0.0071,
1706
- "benchmark": "Arena agent, net improvement",
1707
- "version": "2026-09-27",
1708
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1709
- "date": "2026-09-27",
1710
- "confidence": "adjacent",
1711
- "source": "arena",
1712
- "note": "arena has it at xhigh; carried to this effort"
1713
- },
1714
- {
1715
- "rung": "gpt-5.6-terra#max",
1716
- "dim": "steer",
1717
- "value": 0.0423,
1718
- "benchmark": "Arena agent steerability",
1719
- "version": "2026-09-27",
1720
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1721
- "date": "2026-09-27",
1722
- "confidence": "adjacent",
1723
- "source": "arena",
1724
- "note": "arena has it at xhigh; carried to this effort"
1725
- },
1726
- {
1727
- "rung": "gpt-5.6-terra#max",
1728
- "dim": "frontend",
1729
- "value": 1520.5493,
1730
- "benchmark": "Arena WebDev rating",
1731
- "version": "2026-09-25",
1732
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1733
- "date": "2026-09-25",
1734
- "confidence": "adjacent",
1735
- "source": "arena",
1736
- "note": "arena has it at xhigh; carried to this effort"
1737
- },
1738
- {
1739
- "rung": "gpt-5.6-terra#medium",
1740
- "dim": "agentic",
1741
- "value": 0.0071,
1742
- "benchmark": "Arena agent, net improvement",
1743
- "version": "2026-09-27",
1744
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1745
- "date": "2026-09-27",
1746
- "confidence": "adjacent",
1747
- "source": "arena",
1748
- "note": "arena has it at xhigh; carried to this effort"
1749
- },
1750
- {
1751
- "rung": "gpt-5.6-terra#medium",
1752
- "dim": "steer",
1753
- "value": 0.0423,
1754
- "benchmark": "Arena agent steerability",
1755
- "version": "2026-09-27",
1756
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1757
- "date": "2026-09-27",
1758
- "confidence": "adjacent",
1759
- "source": "arena",
1760
- "note": "arena has it at xhigh; carried to this effort"
1761
- },
1762
- {
1763
- "rung": "gpt-5.6-terra#medium",
1764
- "dim": "frontend",
1765
- "value": 1520.5493,
1766
- "benchmark": "Arena WebDev rating",
1767
- "version": "2026-09-25",
1768
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1769
- "date": "2026-09-25",
1770
- "confidence": "adjacent",
1771
- "source": "arena",
1772
- "note": "arena has it at xhigh; carried to this effort"
1773
- },
1774
- {
1775
- "rung": "gpt-5.6-terra#none",
1776
- "dim": "agentic",
1777
- "value": 0.0071,
1778
- "benchmark": "Arena agent, net improvement",
1779
- "version": "2026-09-27",
1780
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1781
- "date": "2026-09-27",
1782
- "confidence": "adjacent",
1783
- "source": "arena",
1784
- "note": "arena has it at xhigh; carried to this effort"
1785
- },
1786
- {
1787
- "rung": "gpt-5.6-terra#none",
1788
- "dim": "steer",
1789
- "value": 0.0423,
1790
- "benchmark": "Arena agent steerability",
1791
- "version": "2026-09-27",
1792
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1793
- "date": "2026-09-27",
1794
- "confidence": "adjacent",
1795
- "source": "arena",
1796
- "note": "arena has it at xhigh; carried to this effort"
1797
- },
1798
- {
1799
- "rung": "gpt-5.6-terra#none",
1800
- "dim": "frontend",
1801
- "value": 1520.5493,
1802
- "benchmark": "Arena WebDev rating",
1803
- "version": "2026-09-25",
1804
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1805
- "date": "2026-09-25",
1806
- "confidence": "adjacent",
1807
- "source": "arena",
1808
- "note": "arena has it at xhigh; carried to this effort"
1809
- },
1810
- {
1811
- "rung": "gpt-5.6-terra#ultra",
1812
- "dim": "agentic",
1813
- "value": 0.0071,
1814
- "benchmark": "Arena agent, net improvement",
1815
- "version": "2026-09-27",
1816
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1817
- "date": "2026-09-27",
1818
- "confidence": "adjacent",
1819
- "source": "arena",
1820
- "note": "arena has it at xhigh; carried to this effort"
1821
- },
1822
- {
1823
- "rung": "gpt-5.6-terra#ultra",
1824
- "dim": "steer",
1825
- "value": 0.0423,
1826
- "benchmark": "Arena agent steerability",
1827
- "version": "2026-09-27",
1828
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1829
- "date": "2026-09-27",
1830
- "confidence": "adjacent",
1831
- "source": "arena",
1832
- "note": "arena has it at xhigh; carried to this effort"
1833
- },
1834
- {
1835
- "rung": "gpt-5.6-terra#ultra",
1836
- "dim": "frontend",
1837
- "value": 1520.5493,
1838
- "benchmark": "Arena WebDev rating",
1839
- "version": "2026-09-25",
1840
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1841
- "date": "2026-09-25",
1842
- "confidence": "adjacent",
1843
- "source": "arena",
1844
- "note": "arena has it at xhigh; carried to this effort"
1845
- },
1846
- {
1847
- "rung": "gpt-5.6-terra#xhigh",
1848
- "dim": "agentic",
1849
- "value": 0.0071,
1850
- "benchmark": "Arena agent, net improvement",
1851
- "version": "2026-09-27",
1852
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1853
- "date": "2026-09-27",
1854
- "confidence": "measured",
1855
- "source": "arena"
1856
- },
1857
- {
1858
- "rung": "gpt-5.6-terra#xhigh",
1859
- "dim": "steer",
1860
- "value": 0.0423,
1861
- "benchmark": "Arena agent steerability",
1862
- "version": "2026-09-27",
1863
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1864
- "date": "2026-09-27",
1865
- "confidence": "measured",
1866
- "source": "arena"
1867
- },
1868
- {
1869
- "rung": "gpt-5.6-terra#xhigh",
1870
- "dim": "frontend",
1871
- "value": 1520.5493,
1872
- "benchmark": "Arena WebDev rating",
1873
- "version": "2026-09-25",
1874
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1875
- "date": "2026-09-25",
1876
- "confidence": "measured",
1877
- "source": "arena"
1878
- },
1879
- {
1880
- "rung": "gpt-6-astra#high",
1881
- "dim": "repo_code",
1882
- "value": 74,
1883
- "benchmark": "DeepSWE",
1884
- "version": "1.1",
1885
- "url": "https://deepswe.datacurve.ai/",
1886
- "date": "2026-09-22",
1887
- "confidence": "adjacent",
1888
- "note": "DeepSWE has it at xhigh; carried to this effort"
1889
- },
1890
- {
1891
- "rung": "gpt-6-astra#high",
1892
- "dim": "terminal",
1893
- "value": 57.9,
1894
- "benchmark": "Terminal-Bench",
1895
- "version": "4.0",
1896
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
1897
- "date": "2026-09-22",
1898
- "confidence": "secondary",
1899
- "note": "OpenAI-reported, as quoted in the Opus 5.5 system card"
1900
- },
1901
- {
1902
- "rung": "gpt-6-astra#high",
1903
- "dim": "honesty",
1904
- "value": 98.5,
1905
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
1906
- "version": "GPT-6 system card appendix 11.6.4.2",
1907
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
1908
- "date": "2026-09-22",
1909
- "confidence": "adjacent",
1910
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
1911
- },
1912
- {
1913
- "rung": "gpt-6-astra#high",
1914
- "dim": "agentic",
1915
- "value": 0.1031,
1916
- "benchmark": "Arena agent, net improvement",
1917
- "version": "2026-09-27",
1918
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1919
- "date": "2026-09-27",
1920
- "confidence": "adjacent",
1921
- "source": "arena",
1922
- "note": "arena has it at max; carried to this effort"
1923
- },
1924
- {
1925
- "rung": "gpt-6-astra#high",
1926
- "dim": "steer",
1927
- "value": -0.021,
1928
- "benchmark": "Arena agent steerability",
1929
- "version": "2026-09-27",
1930
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1931
- "date": "2026-09-27",
1932
- "confidence": "adjacent",
1933
- "source": "arena",
1934
- "note": "arena has it at max; carried to this effort"
1935
- },
1936
- {
1937
- "rung": "gpt-6-astra#high",
1938
- "dim": "frontend",
1939
- "value": 1791.6526,
1940
- "benchmark": "Arena WebDev rating",
1941
- "version": "2026-09-25",
1942
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1943
- "date": "2026-09-25",
1944
- "confidence": "adjacent",
1945
- "source": "arena",
1946
- "note": "arena has it at max; carried to this effort"
1947
- },
1948
- {
1949
- "rung": "gpt-6-astra#low",
1950
- "dim": "repo_code",
1951
- "value": 74,
1952
- "benchmark": "DeepSWE",
1953
- "version": "1.1",
1954
- "url": "https://deepswe.datacurve.ai/",
1955
- "date": "2026-09-22",
1956
- "confidence": "adjacent",
1957
- "note": "DeepSWE has it at xhigh; carried to this effort"
1958
- },
1959
- {
1960
- "rung": "gpt-6-astra#low",
1961
- "dim": "terminal",
1962
- "value": 57.9,
1963
- "benchmark": "Terminal-Bench",
1964
- "version": "4.0",
1965
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
1966
- "date": "2026-09-22",
1967
- "confidence": "adjacent",
1968
- "note": "Terminal-Bench has it at high; carried to this effort"
1969
- },
1970
- {
1971
- "rung": "gpt-6-astra#low",
1972
- "dim": "honesty",
1973
- "value": 98.5,
1974
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
1975
- "version": "GPT-6 system card appendix 11.6.4.2",
1976
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
1977
- "date": "2026-09-22",
1978
- "confidence": "adjacent",
1979
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
1980
- },
1981
- {
1982
- "rung": "gpt-6-astra#low",
1983
- "dim": "agentic",
1984
- "value": 0.1031,
1985
- "benchmark": "Arena agent, net improvement",
1986
- "version": "2026-09-27",
1987
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
1988
- "date": "2026-09-27",
1989
- "confidence": "adjacent",
1990
- "source": "arena",
1991
- "note": "arena has it at max; carried to this effort"
1992
- },
1993
- {
1994
- "rung": "gpt-6-astra#low",
1995
- "dim": "steer",
1996
- "value": -0.021,
1997
- "benchmark": "Arena agent steerability",
1998
- "version": "2026-09-27",
1999
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2000
- "date": "2026-09-27",
2001
- "confidence": "adjacent",
2002
- "source": "arena",
2003
- "note": "arena has it at max; carried to this effort"
2004
- },
2005
- {
2006
- "rung": "gpt-6-astra#low",
2007
- "dim": "frontend",
2008
- "value": 1791.6526,
2009
- "benchmark": "Arena WebDev rating",
2010
- "version": "2026-09-25",
2011
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2012
- "date": "2026-09-25",
2013
- "confidence": "adjacent",
2014
- "source": "arena",
2015
- "note": "arena has it at max; carried to this effort"
2016
- },
2017
- {
2018
- "rung": "gpt-6-astra#max",
2019
- "dim": "repo_code",
2020
- "value": 74,
2021
- "benchmark": "DeepSWE",
2022
- "version": "1.1",
2023
- "url": "https://deepswe.datacurve.ai/",
2024
- "date": "2026-09-22",
2025
- "confidence": "adjacent",
2026
- "note": "DeepSWE has it at xhigh; carried to this effort"
2027
- },
2028
- {
2029
- "rung": "gpt-6-astra#max",
2030
- "dim": "terminal",
2031
- "value": 57.9,
2032
- "benchmark": "Terminal-Bench",
2033
- "version": "4.0",
2034
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
2035
- "date": "2026-09-22",
2036
- "confidence": "adjacent",
2037
- "note": "Terminal-Bench has it at high; carried to this effort"
2038
- },
2039
- {
2040
- "rung": "gpt-6-astra#max",
2041
- "dim": "honesty",
2042
- "value": 98.5,
2043
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
2044
- "version": "GPT-6 system card appendix 11.6.4.2",
2045
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
2046
- "date": "2026-09-22",
2047
- "confidence": "verified"
2048
- },
2049
- {
2050
- "rung": "gpt-6-astra#max",
2051
- "dim": "agentic",
2052
- "value": 0.1031,
2053
- "benchmark": "Arena agent, net improvement",
2054
- "version": "2026-09-27",
2055
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2056
- "date": "2026-09-27",
2057
- "confidence": "measured",
2058
- "source": "arena"
2059
- },
2060
- {
2061
- "rung": "gpt-6-astra#max",
2062
- "dim": "steer",
2063
- "value": -0.021,
2064
- "benchmark": "Arena agent steerability",
2065
- "version": "2026-09-27",
2066
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2067
- "date": "2026-09-27",
2068
- "confidence": "measured",
2069
- "source": "arena"
2070
- },
2071
- {
2072
- "rung": "gpt-6-astra#max",
2073
- "dim": "frontend",
2074
- "value": 1791.6526,
2075
- "benchmark": "Arena WebDev rating",
2076
- "version": "2026-09-25",
2077
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2078
- "date": "2026-09-25",
2079
- "confidence": "measured",
2080
- "source": "arena"
2081
- },
2082
- {
2083
- "rung": "gpt-6-astra#medium",
2084
- "dim": "repo_code",
2085
- "value": 74,
2086
- "benchmark": "DeepSWE",
2087
- "version": "1.1",
2088
- "url": "https://deepswe.datacurve.ai/",
2089
- "date": "2026-09-22",
2090
- "confidence": "adjacent",
2091
- "note": "DeepSWE has it at xhigh; carried to this effort"
2092
- },
2093
- {
2094
- "rung": "gpt-6-astra#medium",
2095
- "dim": "terminal",
2096
- "value": 57.9,
2097
- "benchmark": "Terminal-Bench",
2098
- "version": "4.0",
2099
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
2100
- "date": "2026-09-22",
2101
- "confidence": "adjacent",
2102
- "note": "Terminal-Bench has it at high; carried to this effort"
2103
- },
2104
- {
2105
- "rung": "gpt-6-astra#medium",
2106
- "dim": "honesty",
2107
- "value": 98.5,
2108
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
2109
- "version": "GPT-6 system card appendix 11.6.4.2",
2110
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
2111
- "date": "2026-09-22",
2112
- "confidence": "adjacent",
2113
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
2114
- },
2115
- {
2116
- "rung": "gpt-6-astra#medium",
2117
- "dim": "agentic",
2118
- "value": 0.1031,
2119
- "benchmark": "Arena agent, net improvement",
2120
- "version": "2026-09-27",
2121
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2122
- "date": "2026-09-27",
2123
- "confidence": "adjacent",
2124
- "source": "arena",
2125
- "note": "arena has it at max; carried to this effort"
2126
- },
2127
- {
2128
- "rung": "gpt-6-astra#medium",
2129
- "dim": "steer",
2130
- "value": -0.021,
2131
- "benchmark": "Arena agent steerability",
2132
- "version": "2026-09-27",
2133
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2134
- "date": "2026-09-27",
2135
- "confidence": "adjacent",
2136
- "source": "arena",
2137
- "note": "arena has it at max; carried to this effort"
2138
- },
2139
- {
2140
- "rung": "gpt-6-astra#medium",
2141
- "dim": "frontend",
2142
- "value": 1791.6526,
2143
- "benchmark": "Arena WebDev rating",
2144
- "version": "2026-09-25",
2145
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2146
- "date": "2026-09-25",
2147
- "confidence": "adjacent",
2148
- "source": "arena",
2149
- "note": "arena has it at max; carried to this effort"
2150
- },
2151
- {
2152
- "rung": "gpt-6-astra#ultra",
2153
- "dim": "repo_code",
2154
- "value": 74,
2155
- "benchmark": "DeepSWE",
2156
- "version": "1.1",
2157
- "url": "https://deepswe.datacurve.ai/",
2158
- "date": "2026-09-22",
2159
- "confidence": "adjacent",
2160
- "note": "DeepSWE has it at xhigh; carried to this effort"
2161
- },
2162
- {
2163
- "rung": "gpt-6-astra#ultra",
2164
- "dim": "terminal",
2165
- "value": 57.9,
2166
- "benchmark": "Terminal-Bench",
2167
- "version": "4.0",
2168
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
2169
- "date": "2026-09-22",
2170
- "confidence": "adjacent",
2171
- "note": "Terminal-Bench has it at high; carried to this effort"
2172
- },
2173
- {
2174
- "rung": "gpt-6-astra#ultra",
2175
- "dim": "honesty",
2176
- "value": 98.5,
2177
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
2178
- "version": "GPT-6 system card appendix 11.6.4.2",
2179
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
2180
- "date": "2026-09-22",
2181
- "confidence": "adjacent",
2182
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
2183
- },
2184
- {
2185
- "rung": "gpt-6-astra#ultra",
2186
- "dim": "agentic",
2187
- "value": 0.1031,
2188
- "benchmark": "Arena agent, net improvement",
2189
- "version": "2026-09-27",
2190
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2191
- "date": "2026-09-27",
2192
- "confidence": "adjacent",
2193
- "source": "arena",
2194
- "note": "arena has it at max; carried to this effort"
2195
- },
2196
- {
2197
- "rung": "gpt-6-astra#ultra",
2198
- "dim": "steer",
2199
- "value": -0.021,
2200
- "benchmark": "Arena agent steerability",
2201
- "version": "2026-09-27",
2202
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2203
- "date": "2026-09-27",
2204
- "confidence": "adjacent",
2205
- "source": "arena",
2206
- "note": "arena has it at max; carried to this effort"
2207
- },
2208
- {
2209
- "rung": "gpt-6-astra#ultra",
2210
- "dim": "frontend",
2211
- "value": 1791.6526,
2212
- "benchmark": "Arena WebDev rating",
2213
- "version": "2026-09-25",
2214
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2215
- "date": "2026-09-25",
2216
- "confidence": "adjacent",
2217
- "source": "arena",
2218
- "note": "arena has it at max; carried to this effort"
2219
- },
2220
- {
2221
- "rung": "gpt-6-astra#xhigh",
2222
- "dim": "repo_code",
2223
- "value": 74,
2224
- "benchmark": "DeepSWE",
2225
- "version": "1.1",
2226
- "url": "https://deepswe.datacurve.ai/",
2227
- "date": "2026-09-22",
2228
- "confidence": "verified"
2229
- },
2230
- {
2231
- "rung": "gpt-6-astra#xhigh",
2232
- "dim": "terminal",
2233
- "value": 57.9,
2234
- "benchmark": "Terminal-Bench",
2235
- "version": "4.0",
2236
- "url": "https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf",
2237
- "date": "2026-09-22",
2238
- "confidence": "adjacent",
2239
- "note": "Terminal-Bench has it at high; carried to this effort"
2240
- },
2241
- {
2242
- "rung": "gpt-6-astra#xhigh",
2243
- "dim": "honesty",
2244
- "value": 98.5,
2245
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
2246
- "version": "GPT-6 system card appendix 11.6.4.2",
2247
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
2248
- "date": "2026-09-22",
2249
- "confidence": "adjacent",
2250
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
2251
- },
2252
- {
2253
- "rung": "gpt-6-astra#xhigh",
2254
- "dim": "agentic",
2255
- "value": 0.1031,
2256
- "benchmark": "Arena agent, net improvement",
2257
- "version": "2026-09-27",
2258
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2259
- "date": "2026-09-27",
2260
- "confidence": "adjacent",
2261
- "source": "arena",
2262
- "note": "arena has it at max; carried to this effort"
2263
- },
2264
- {
2265
- "rung": "gpt-6-astra#xhigh",
2266
- "dim": "steer",
2267
- "value": -0.021,
2268
- "benchmark": "Arena agent steerability",
2269
- "version": "2026-09-27",
2270
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2271
- "date": "2026-09-27",
2272
- "confidence": "adjacent",
2273
- "source": "arena",
2274
- "note": "arena has it at max; carried to this effort"
2275
- },
2276
- {
2277
- "rung": "gpt-6-astra#xhigh",
2278
- "dim": "frontend",
2279
- "value": 1791.6526,
2280
- "benchmark": "Arena WebDev rating",
2281
- "version": "2026-09-25",
2282
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2283
- "date": "2026-09-25",
2284
- "confidence": "adjacent",
2285
- "source": "arena",
2286
- "note": "arena has it at max; carried to this effort"
2287
- },
2288
- {
2289
- "rung": "gpt-6-luna#high",
2290
- "dim": "repo_code",
2291
- "value": 66.6,
2292
- "benchmark": "DeepSWE",
2293
- "version": "1.1",
2294
- "url": "https://kingy.ai/blog/gpt-6-sol-luna-specs-benchmarks-pricing-comparison/",
2295
- "date": "2026-09-22",
2296
- "confidence": "adjacent",
2297
- "note": "DeepSWE has it at max; carried to this effort"
2298
- },
2299
- {
2300
- "rung": "gpt-6-luna#high",
2301
- "dim": "terminal",
2302
- "value": 13,
2303
- "benchmark": "Terminal-Bench",
2304
- "version": "4.0",
2305
- "url": "https://artificialanalysis.ai/articles/gpt-6-sol-and-luna-push-the-cost-efficiency-frontier",
2306
- "date": "2026-09-22",
2307
- "confidence": "adjacent",
2308
- "note": "Terminal-Bench has it at max; carried to this effort"
2309
- },
2310
- {
2311
- "rung": "gpt-6-luna#high",
2312
- "dim": "honesty",
2313
- "value": 71.3,
2314
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
2315
- "version": "GPT-6 system card appendix 11.6.4.2",
2316
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
2317
- "date": "2026-09-22",
2318
- "confidence": "adjacent",
2319
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
2320
- },
2321
- {
2322
- "rung": "gpt-6-luna#high",
2323
- "dim": "frontend",
2324
- "value": 1592.6467,
2325
- "benchmark": "Arena WebDev rating",
2326
- "version": "2026-09-25",
2327
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2328
- "date": "2026-09-25",
2329
- "confidence": "adjacent",
2330
- "source": "arena",
2331
- "note": "arena has it at max; carried to this effort"
2332
- },
2333
- {
2334
- "rung": "gpt-6-luna#low",
2335
- "dim": "repo_code",
2336
- "value": 66.6,
2337
- "benchmark": "DeepSWE",
2338
- "version": "1.1",
2339
- "url": "https://kingy.ai/blog/gpt-6-sol-luna-specs-benchmarks-pricing-comparison/",
2340
- "date": "2026-09-22",
2341
- "confidence": "adjacent",
2342
- "note": "DeepSWE has it at max; carried to this effort"
2343
- },
2344
- {
2345
- "rung": "gpt-6-luna#low",
2346
- "dim": "terminal",
2347
- "value": 13,
2348
- "benchmark": "Terminal-Bench",
2349
- "version": "4.0",
2350
- "url": "https://artificialanalysis.ai/articles/gpt-6-sol-and-luna-push-the-cost-efficiency-frontier",
2351
- "date": "2026-09-22",
2352
- "confidence": "adjacent",
2353
- "note": "Terminal-Bench has it at max; carried to this effort"
2354
- },
2355
- {
2356
- "rung": "gpt-6-luna#low",
2357
- "dim": "honesty",
2358
- "value": 71.3,
2359
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
2360
- "version": "GPT-6 system card appendix 11.6.4.2",
2361
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
2362
- "date": "2026-09-22",
2363
- "confidence": "adjacent",
2364
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
2365
- },
2366
- {
2367
- "rung": "gpt-6-luna#low",
2368
- "dim": "frontend",
2369
- "value": 1592.6467,
2370
- "benchmark": "Arena WebDev rating",
2371
- "version": "2026-09-25",
2372
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2373
- "date": "2026-09-25",
2374
- "confidence": "adjacent",
2375
- "source": "arena",
2376
- "note": "arena has it at max; carried to this effort"
2377
- },
2378
- {
2379
- "rung": "gpt-6-luna#max",
2380
- "dim": "repo_code",
2381
- "value": 66.6,
2382
- "benchmark": "DeepSWE",
2383
- "version": "1.1",
2384
- "url": "https://kingy.ai/blog/gpt-6-sol-luna-specs-benchmarks-pricing-comparison/",
2385
- "date": "2026-09-22",
2386
- "confidence": "secondary"
2387
- },
2388
- {
2389
- "rung": "gpt-6-luna#max",
2390
- "dim": "terminal",
2391
- "value": 13,
2392
- "benchmark": "Terminal-Bench",
2393
- "version": "4.0",
2394
- "url": "https://artificialanalysis.ai/articles/gpt-6-sol-and-luna-push-the-cost-efficiency-frontier",
2395
- "date": "2026-09-22",
2396
- "confidence": "secondary"
2397
- },
2398
- {
2399
- "rung": "gpt-6-luna#max",
2400
- "dim": "honesty",
2401
- "value": 71.3,
2402
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
2403
- "version": "GPT-6 system card appendix 11.6.4.2",
2404
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
2405
- "date": "2026-09-22",
2406
- "confidence": "verified"
2407
- },
2408
- {
2409
- "rung": "gpt-6-luna#max",
2410
- "dim": "frontend",
2411
- "value": 1592.6467,
2412
- "benchmark": "Arena WebDev rating",
2413
- "version": "2026-09-25",
2414
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2415
- "date": "2026-09-25",
2416
- "confidence": "measured",
2417
- "source": "arena"
2418
- },
2419
- {
2420
- "rung": "gpt-6-luna#medium",
2421
- "dim": "repo_code",
2422
- "value": 66.6,
2423
- "benchmark": "DeepSWE",
2424
- "version": "1.1",
2425
- "url": "https://kingy.ai/blog/gpt-6-sol-luna-specs-benchmarks-pricing-comparison/",
2426
- "date": "2026-09-22",
2427
- "confidence": "adjacent",
2428
- "note": "DeepSWE has it at max; carried to this effort"
2429
- },
2430
- {
2431
- "rung": "gpt-6-luna#medium",
2432
- "dim": "terminal",
2433
- "value": 13,
2434
- "benchmark": "Terminal-Bench",
2435
- "version": "4.0",
2436
- "url": "https://artificialanalysis.ai/articles/gpt-6-sol-and-luna-push-the-cost-efficiency-frontier",
2437
- "date": "2026-09-22",
2438
- "confidence": "adjacent",
2439
- "note": "Terminal-Bench has it at max; carried to this effort"
2440
- },
2441
- {
2442
- "rung": "gpt-6-luna#medium",
2443
- "dim": "honesty",
2444
- "value": 71.3,
2445
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
2446
- "version": "GPT-6 system card appendix 11.6.4.2",
2447
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
2448
- "date": "2026-09-22",
2449
- "confidence": "adjacent",
2450
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
2451
- },
2452
- {
2453
- "rung": "gpt-6-luna#medium",
2454
- "dim": "frontend",
2455
- "value": 1592.6467,
2456
- "benchmark": "Arena WebDev rating",
2457
- "version": "2026-09-25",
2458
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2459
- "date": "2026-09-25",
2460
- "confidence": "adjacent",
2461
- "source": "arena",
2462
- "note": "arena has it at max; carried to this effort"
2463
- },
2464
- {
2465
- "rung": "gpt-6-luna#none",
2466
- "dim": "repo_code",
2467
- "value": 66.6,
2468
- "benchmark": "DeepSWE",
2469
- "version": "1.1",
2470
- "url": "https://kingy.ai/blog/gpt-6-sol-luna-specs-benchmarks-pricing-comparison/",
2471
- "date": "2026-09-22",
2472
- "confidence": "adjacent",
2473
- "note": "DeepSWE has it at max; carried to this effort"
2474
- },
2475
- {
2476
- "rung": "gpt-6-luna#none",
2477
- "dim": "terminal",
2478
- "value": 13,
2479
- "benchmark": "Terminal-Bench",
2480
- "version": "4.0",
2481
- "url": "https://artificialanalysis.ai/articles/gpt-6-sol-and-luna-push-the-cost-efficiency-frontier",
2482
- "date": "2026-09-22",
2483
- "confidence": "adjacent",
2484
- "note": "Terminal-Bench has it at max; carried to this effort"
2485
- },
2486
- {
2487
- "rung": "gpt-6-luna#none",
2488
- "dim": "honesty",
2489
- "value": 71.3,
2490
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
2491
- "version": "GPT-6 system card appendix 11.6.4.2",
2492
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
2493
- "date": "2026-09-22",
2494
- "confidence": "adjacent",
2495
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
2496
- },
2497
- {
2498
- "rung": "gpt-6-luna#none",
2499
- "dim": "frontend",
2500
- "value": 1592.6467,
2501
- "benchmark": "Arena WebDev rating",
2502
- "version": "2026-09-25",
2503
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2504
- "date": "2026-09-25",
2505
- "confidence": "adjacent",
2506
- "source": "arena",
2507
- "note": "arena has it at max; carried to this effort"
2508
- },
2509
- {
2510
- "rung": "gpt-6-luna#xhigh",
2511
- "dim": "repo_code",
2512
- "value": 66.6,
2513
- "benchmark": "DeepSWE",
2514
- "version": "1.1",
2515
- "url": "https://kingy.ai/blog/gpt-6-sol-luna-specs-benchmarks-pricing-comparison/",
2516
- "date": "2026-09-22",
2517
- "confidence": "adjacent",
2518
- "note": "DeepSWE has it at max; carried to this effort"
2519
- },
2520
- {
2521
- "rung": "gpt-6-luna#xhigh",
2522
- "dim": "terminal",
2523
- "value": 13,
2524
- "benchmark": "Terminal-Bench",
2525
- "version": "4.0",
2526
- "url": "https://artificialanalysis.ai/articles/gpt-6-sol-and-luna-push-the-cost-efficiency-frontier",
2527
- "date": "2026-09-22",
2528
- "confidence": "adjacent",
2529
- "note": "Terminal-Bench has it at max; carried to this effort"
2530
- },
2531
- {
2532
- "rung": "gpt-6-luna#xhigh",
2533
- "dim": "honesty",
2534
- "value": 71.3,
2535
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
2536
- "version": "GPT-6 system card appendix 11.6.4.2",
2537
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
2538
- "date": "2026-09-22",
2539
- "confidence": "adjacent",
2540
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
2541
- },
2542
- {
2543
- "rung": "gpt-6-luna#xhigh",
2544
- "dim": "frontend",
2545
- "value": 1592.6467,
2546
- "benchmark": "Arena WebDev rating",
2547
- "version": "2026-09-25",
2548
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2549
- "date": "2026-09-25",
2550
- "confidence": "adjacent",
2551
- "source": "arena",
2552
- "note": "arena has it at max; carried to this effort"
2553
- },
2554
- {
2555
- "rung": "gpt-6-sol#high",
2556
- "dim": "repo_code",
2557
- "value": 65.3,
2558
- "benchmark": "DeepSWE",
2559
- "version": "1.1",
2560
- "url": "https://kingy.ai/blog/gpt-6-sol-luna-specs-benchmarks-pricing-comparison/",
2561
- "date": "2026-09-22",
2562
- "confidence": "secondary"
2563
- },
2564
- {
2565
- "rung": "gpt-6-sol#high",
2566
- "dim": "terminal",
2567
- "value": 43,
2568
- "benchmark": "Terminal-Bench",
2569
- "version": "4.0",
2570
- "url": "https://artificialanalysis.ai/articles/gpt-6-sol-and-luna-push-the-cost-efficiency-frontier",
2571
- "date": "2026-09-22",
2572
- "confidence": "adjacent",
2573
- "note": "Terminal-Bench has it at max; carried to this effort"
2574
- },
2575
- {
2576
- "rung": "gpt-6-sol#high",
2577
- "dim": "honesty",
2578
- "value": 95.1,
2579
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
2580
- "version": "GPT-6 system card appendix 11.6.4.2",
2581
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
2582
- "date": "2026-09-22",
2583
- "confidence": "adjacent",
2584
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
2585
- },
2586
- {
2587
- "rung": "gpt-6-sol#high",
2588
- "dim": "agentic",
2589
- "value": 0.0818,
2590
- "benchmark": "Arena agent, net improvement",
2591
- "version": "2026-09-27",
2592
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2593
- "date": "2026-09-27",
2594
- "confidence": "adjacent",
2595
- "source": "arena",
2596
- "note": "arena has it at max; carried to this effort"
2597
- },
2598
- {
2599
- "rung": "gpt-6-sol#high",
2600
- "dim": "steer",
2601
- "value": 0.1335,
2602
- "benchmark": "Arena agent steerability",
2603
- "version": "2026-09-27",
2604
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2605
- "date": "2026-09-27",
2606
- "confidence": "adjacent",
2607
- "source": "arena",
2608
- "note": "arena has it at max; carried to this effort"
2609
- },
2610
- {
2611
- "rung": "gpt-6-sol#high",
2612
- "dim": "frontend",
2613
- "value": 1680.9725,
2614
- "benchmark": "Arena WebDev rating",
2615
- "version": "2026-09-25",
2616
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2617
- "date": "2026-09-25",
2618
- "confidence": "adjacent",
2619
- "source": "arena",
2620
- "note": "arena has it at max; carried to this effort"
2621
- },
2622
- {
2623
- "rung": "gpt-6-sol#low",
2624
- "dim": "repo_code",
2625
- "value": 37.2,
2626
- "benchmark": "DeepSWE",
2627
- "version": "1.1",
2628
- "url": "https://kingy.ai/blog/gpt-6-sol-luna-specs-benchmarks-pricing-comparison/",
2629
- "date": "2026-09-22",
2630
- "confidence": "secondary"
2631
- },
2632
- {
2633
- "rung": "gpt-6-sol#low",
2634
- "dim": "terminal",
2635
- "value": 43,
2636
- "benchmark": "Terminal-Bench",
2637
- "version": "4.0",
2638
- "url": "https://artificialanalysis.ai/articles/gpt-6-sol-and-luna-push-the-cost-efficiency-frontier",
2639
- "date": "2026-09-22",
2640
- "confidence": "adjacent",
2641
- "note": "Terminal-Bench has it at max; carried to this effort"
2642
- },
2643
- {
2644
- "rung": "gpt-6-sol#low",
2645
- "dim": "honesty",
2646
- "value": 95.1,
2647
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
2648
- "version": "GPT-6 system card appendix 11.6.4.2",
2649
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
2650
- "date": "2026-09-22",
2651
- "confidence": "adjacent",
2652
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
2653
- },
2654
- {
2655
- "rung": "gpt-6-sol#low",
2656
- "dim": "agentic",
2657
- "value": 0.0818,
2658
- "benchmark": "Arena agent, net improvement",
2659
- "version": "2026-09-27",
2660
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2661
- "date": "2026-09-27",
2662
- "confidence": "adjacent",
2663
- "source": "arena",
2664
- "note": "arena has it at max; carried to this effort"
2665
- },
2666
- {
2667
- "rung": "gpt-6-sol#low",
2668
- "dim": "steer",
2669
- "value": 0.1335,
2670
- "benchmark": "Arena agent steerability",
2671
- "version": "2026-09-27",
2672
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2673
- "date": "2026-09-27",
2674
- "confidence": "adjacent",
2675
- "source": "arena",
2676
- "note": "arena has it at max; carried to this effort"
2677
- },
2678
- {
2679
- "rung": "gpt-6-sol#low",
2680
- "dim": "frontend",
2681
- "value": 1680.9725,
2682
- "benchmark": "Arena WebDev rating",
2683
- "version": "2026-09-25",
2684
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2685
- "date": "2026-09-25",
2686
- "confidence": "adjacent",
2687
- "source": "arena",
2688
- "note": "arena has it at max; carried to this effort"
2689
- },
2690
- {
2691
- "rung": "gpt-6-sol#max",
2692
- "dim": "repo_code",
2693
- "value": 68.8,
2694
- "benchmark": "DeepSWE",
2695
- "version": "1.1",
2696
- "url": "https://x.com/OpenAIDevs/status/2102461464279912495",
2697
- "date": "2026-09-22",
2698
- "confidence": "verified"
2699
- },
2700
- {
2701
- "rung": "gpt-6-sol#max",
2702
- "dim": "terminal",
2703
- "value": 43,
2704
- "benchmark": "Terminal-Bench",
2705
- "version": "4.0",
2706
- "url": "https://artificialanalysis.ai/articles/gpt-6-sol-and-luna-push-the-cost-efficiency-frontier",
2707
- "date": "2026-09-22",
2708
- "confidence": "secondary"
2709
- },
2710
- {
2711
- "rung": "gpt-6-sol#max",
2712
- "dim": "honesty",
2713
- "value": 95.1,
2714
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
2715
- "version": "GPT-6 system card appendix 11.6.4.2",
2716
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
2717
- "date": "2026-09-22",
2718
- "confidence": "verified"
2719
- },
2720
- {
2721
- "rung": "gpt-6-sol#max",
2722
- "dim": "agentic",
2723
- "value": 0.0818,
2724
- "benchmark": "Arena agent, net improvement",
2725
- "version": "2026-09-27",
2726
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2727
- "date": "2026-09-27",
2728
- "confidence": "measured",
2729
- "source": "arena"
2730
- },
2731
- {
2732
- "rung": "gpt-6-sol#max",
2733
- "dim": "steer",
2734
- "value": 0.1335,
2735
- "benchmark": "Arena agent steerability",
2736
- "version": "2026-09-27",
2737
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2738
- "date": "2026-09-27",
2739
- "confidence": "measured",
2740
- "source": "arena"
2741
- },
2742
- {
2743
- "rung": "gpt-6-sol#max",
2744
- "dim": "frontend",
2745
- "value": 1680.9725,
2746
- "benchmark": "Arena WebDev rating",
2747
- "version": "2026-09-25",
2748
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2749
- "date": "2026-09-25",
2750
- "confidence": "measured",
2751
- "source": "arena"
2752
- },
2753
- {
2754
- "rung": "gpt-6-sol#medium",
2755
- "dim": "repo_code",
2756
- "value": 56.6,
2757
- "benchmark": "DeepSWE",
2758
- "version": "1.1",
2759
- "url": "https://kingy.ai/blog/gpt-6-sol-luna-specs-benchmarks-pricing-comparison/",
2760
- "date": "2026-09-22",
2761
- "confidence": "secondary"
2762
- },
2763
- {
2764
- "rung": "gpt-6-sol#medium",
2765
- "dim": "terminal",
2766
- "value": 43,
2767
- "benchmark": "Terminal-Bench",
2768
- "version": "4.0",
2769
- "url": "https://artificialanalysis.ai/articles/gpt-6-sol-and-luna-push-the-cost-efficiency-frontier",
2770
- "date": "2026-09-22",
2771
- "confidence": "adjacent",
2772
- "note": "Terminal-Bench has it at max; carried to this effort"
2773
- },
2774
- {
2775
- "rung": "gpt-6-sol#medium",
2776
- "dim": "honesty",
2777
- "value": 95.1,
2778
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
2779
- "version": "GPT-6 system card appendix 11.6.4.2",
2780
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
2781
- "date": "2026-09-22",
2782
- "confidence": "adjacent",
2783
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
2784
- },
2785
- {
2786
- "rung": "gpt-6-sol#medium",
2787
- "dim": "agentic",
2788
- "value": 0.0818,
2789
- "benchmark": "Arena agent, net improvement",
2790
- "version": "2026-09-27",
2791
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2792
- "date": "2026-09-27",
2793
- "confidence": "adjacent",
2794
- "source": "arena",
2795
- "note": "arena has it at max; carried to this effort"
2796
- },
2797
- {
2798
- "rung": "gpt-6-sol#medium",
2799
- "dim": "steer",
2800
- "value": 0.1335,
2801
- "benchmark": "Arena agent steerability",
2802
- "version": "2026-09-27",
2803
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2804
- "date": "2026-09-27",
2805
- "confidence": "adjacent",
2806
- "source": "arena",
2807
- "note": "arena has it at max; carried to this effort"
2808
- },
2809
- {
2810
- "rung": "gpt-6-sol#medium",
2811
- "dim": "frontend",
2812
- "value": 1680.9725,
2813
- "benchmark": "Arena WebDev rating",
2814
- "version": "2026-09-25",
2815
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2816
- "date": "2026-09-25",
2817
- "confidence": "adjacent",
2818
- "source": "arena",
2819
- "note": "arena has it at max; carried to this effort"
2820
- },
2821
- {
2822
- "rung": "gpt-6-sol#none",
2823
- "dim": "repo_code",
2824
- "value": 37.2,
2825
- "benchmark": "DeepSWE",
2826
- "version": "1.1",
2827
- "url": "https://kingy.ai/blog/gpt-6-sol-luna-specs-benchmarks-pricing-comparison/",
2828
- "date": "2026-09-22",
2829
- "confidence": "adjacent",
2830
- "note": "DeepSWE has it at low; carried to this effort"
2831
- },
2832
- {
2833
- "rung": "gpt-6-sol#none",
2834
- "dim": "terminal",
2835
- "value": 43,
2836
- "benchmark": "Terminal-Bench",
2837
- "version": "4.0",
2838
- "url": "https://artificialanalysis.ai/articles/gpt-6-sol-and-luna-push-the-cost-efficiency-frontier",
2839
- "date": "2026-09-22",
2840
- "confidence": "adjacent",
2841
- "note": "Terminal-Bench has it at max; carried to this effort"
2842
- },
2843
- {
2844
- "rung": "gpt-6-sol#none",
2845
- "dim": "honesty",
2846
- "value": 95.1,
2847
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
2848
- "version": "GPT-6 system card appendix 11.6.4.2",
2849
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
2850
- "date": "2026-09-22",
2851
- "confidence": "adjacent",
2852
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
2853
- },
2854
- {
2855
- "rung": "gpt-6-sol#none",
2856
- "dim": "agentic",
2857
- "value": 0.0818,
2858
- "benchmark": "Arena agent, net improvement",
2859
- "version": "2026-09-27",
2860
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2861
- "date": "2026-09-27",
2862
- "confidence": "adjacent",
2863
- "source": "arena",
2864
- "note": "arena has it at max; carried to this effort"
2865
- },
2866
- {
2867
- "rung": "gpt-6-sol#none",
2868
- "dim": "steer",
2869
- "value": 0.1335,
2870
- "benchmark": "Arena agent steerability",
2871
- "version": "2026-09-27",
2872
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2873
- "date": "2026-09-27",
2874
- "confidence": "adjacent",
2875
- "source": "arena",
2876
- "note": "arena has it at max; carried to this effort"
2877
- },
2878
- {
2879
- "rung": "gpt-6-sol#none",
2880
- "dim": "frontend",
2881
- "value": 1680.9725,
2882
- "benchmark": "Arena WebDev rating",
2883
- "version": "2026-09-25",
2884
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2885
- "date": "2026-09-25",
2886
- "confidence": "adjacent",
2887
- "source": "arena",
2888
- "note": "arena has it at max; carried to this effort"
2889
- },
2890
- {
2891
- "rung": "gpt-6-sol#ultra",
2892
- "dim": "repo_code",
2893
- "value": 68.8,
2894
- "benchmark": "DeepSWE",
2895
- "version": "1.1",
2896
- "url": "https://x.com/OpenAIDevs/status/2102461464279912495",
2897
- "date": "2026-09-22",
2898
- "confidence": "adjacent",
2899
- "note": "DeepSWE has it at max; carried to this effort"
2900
- },
2901
- {
2902
- "rung": "gpt-6-sol#ultra",
2903
- "dim": "terminal",
2904
- "value": 43,
2905
- "benchmark": "Terminal-Bench",
2906
- "version": "4.0",
2907
- "url": "https://artificialanalysis.ai/articles/gpt-6-sol-and-luna-push-the-cost-efficiency-frontier",
2908
- "date": "2026-09-22",
2909
- "confidence": "adjacent",
2910
- "note": "Terminal-Bench has it at max; carried to this effort"
2911
- },
2912
- {
2913
- "rung": "gpt-6-sol#ultra",
2914
- "dim": "honesty",
2915
- "value": 95.1,
2916
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
2917
- "version": "GPT-6 system card appendix 11.6.4.2",
2918
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
2919
- "date": "2026-09-22",
2920
- "confidence": "adjacent",
2921
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
2922
- },
2923
- {
2924
- "rung": "gpt-6-sol#ultra",
2925
- "dim": "agentic",
2926
- "value": 0.0818,
2927
- "benchmark": "Arena agent, net improvement",
2928
- "version": "2026-09-27",
2929
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2930
- "date": "2026-09-27",
2931
- "confidence": "adjacent",
2932
- "source": "arena",
2933
- "note": "arena has it at max; carried to this effort"
2934
- },
2935
- {
2936
- "rung": "gpt-6-sol#ultra",
2937
- "dim": "steer",
2938
- "value": 0.1335,
2939
- "benchmark": "Arena agent steerability",
2940
- "version": "2026-09-27",
2941
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2942
- "date": "2026-09-27",
2943
- "confidence": "adjacent",
2944
- "source": "arena",
2945
- "note": "arena has it at max; carried to this effort"
2946
- },
2947
- {
2948
- "rung": "gpt-6-sol#ultra",
2949
- "dim": "frontend",
2950
- "value": 1680.9725,
2951
- "benchmark": "Arena WebDev rating",
2952
- "version": "2026-09-25",
2953
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2954
- "date": "2026-09-25",
2955
- "confidence": "adjacent",
2956
- "source": "arena",
2957
- "note": "arena has it at max; carried to this effort"
2958
- },
2959
- {
2960
- "rung": "gpt-6-sol#xhigh",
2961
- "dim": "repo_code",
2962
- "value": 66.6,
2963
- "benchmark": "DeepSWE",
2964
- "version": "1.1",
2965
- "url": "https://kingy.ai/blog/gpt-6-sol-luna-specs-benchmarks-pricing-comparison/",
2966
- "date": "2026-09-22",
2967
- "confidence": "secondary"
2968
- },
2969
- {
2970
- "rung": "gpt-6-sol#xhigh",
2971
- "dim": "terminal",
2972
- "value": 43,
2973
- "benchmark": "Terminal-Bench",
2974
- "version": "4.0",
2975
- "url": "https://artificialanalysis.ai/articles/gpt-6-sol-and-luna-push-the-cost-efficiency-frontier",
2976
- "date": "2026-09-22",
2977
- "confidence": "adjacent",
2978
- "note": "Terminal-Bench has it at max; carried to this effort"
2979
- },
2980
- {
2981
- "rung": "gpt-6-sol#xhigh",
2982
- "dim": "honesty",
2983
- "value": 95.1,
2984
- "benchmark": "OpenAI Broken Search Tool, 100 minus the failure rate",
2985
- "version": "GPT-6 system card appendix 11.6.4.2",
2986
- "url": "https://deploymentsafety.openai.com/gpt-6-astra/sec:appendix-sol-luna",
2987
- "date": "2026-09-22",
2988
- "confidence": "adjacent",
2989
- "note": "OpenAI Broken Search Tool, 100 minus the failure rate has it at max; carried to this effort"
2990
- },
2991
- {
2992
- "rung": "gpt-6-sol#xhigh",
2993
- "dim": "agentic",
2994
- "value": 0.0818,
2995
- "benchmark": "Arena agent, net improvement",
2996
- "version": "2026-09-27",
2997
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
2998
- "date": "2026-09-27",
2999
- "confidence": "adjacent",
3000
- "source": "arena",
3001
- "note": "arena has it at max; carried to this effort"
3002
- },
3003
- {
3004
- "rung": "gpt-6-sol#xhigh",
3005
- "dim": "steer",
3006
- "value": 0.1335,
3007
- "benchmark": "Arena agent steerability",
3008
- "version": "2026-09-27",
3009
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
3010
- "date": "2026-09-27",
3011
- "confidence": "adjacent",
3012
- "source": "arena",
3013
- "note": "arena has it at max; carried to this effort"
3014
- },
3015
- {
3016
- "rung": "gpt-6-sol#xhigh",
3017
- "dim": "frontend",
3018
- "value": 1680.9725,
3019
- "benchmark": "Arena WebDev rating",
3020
- "version": "2026-09-25",
3021
- "url": "https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset",
3022
- "date": "2026-09-25",
3023
- "confidence": "adjacent",
3024
- "source": "arena",
3025
- "note": "arena has it at max; carried to this effort"
3026
- }
3027
- ],
3028
- "treatLike": {
3029
- "opencode-go/kimi-k3#max": {
3030
- "like": "gpt-6-sol#medium",
3031
- "note": "the default profile's OpenCode Go stand-in for Sol; no DeepSWE 1.1, Terminal-Bench 4.0 or honesty score is published for it"
3032
- },
3033
- "gpt-6-luna#none": {
3034
- "like": "gpt-5.6-luna#none",
3035
- "note": "no Arena agent score is published for GPT-6 Luna; GPT-5.6 Luna at the same effort stands in for agentic and steer (the nearest stand-in, spec 1.2 §6.3)"
3036
- },
3037
- "gpt-6-luna#low": {
3038
- "like": "gpt-5.6-luna#low",
3039
- "note": "no Arena agent score is published for GPT-6 Luna; GPT-5.6 Luna at the same effort stands in for agentic and steer (the nearest stand-in, spec 1.2 §6.3)"
3040
- },
3041
- "gpt-6-luna#medium": {
3042
- "like": "gpt-5.6-luna#medium",
3043
- "note": "no Arena agent score is published for GPT-6 Luna; GPT-5.6 Luna at the same effort stands in for agentic and steer (the nearest stand-in, spec 1.2 §6.3)"
3044
- },
3045
- "gpt-6-luna#high": {
3046
- "like": "gpt-5.6-luna#high",
3047
- "note": "no Arena agent score is published for GPT-6 Luna; GPT-5.6 Luna at the same effort stands in for agentic and steer (the nearest stand-in, spec 1.2 §6.3)"
3048
- },
3049
- "gpt-6-luna#xhigh": {
3050
- "like": "gpt-5.6-luna#xhigh",
3051
- "note": "no Arena agent score is published for GPT-6 Luna; GPT-5.6 Luna at the same effort stands in for agentic and steer (the nearest stand-in, spec 1.2 §6.3)"
3052
- },
3053
- "gpt-6-luna#max": {
3054
- "like": "gpt-5.6-luna#max",
3055
- "note": "no Arena agent score is published for GPT-6 Luna; GPT-5.6 Luna at the same effort stands in for agentic and steer (the nearest stand-in, spec 1.2 §6.3)"
3056
- },
3057
- "gpt-6.1-sol#low": {
3058
- "like": "gpt-6-sol#low",
3059
- "note": "no DeepSWE 1.1 or Arena score is published for GPT-6.1 Sol yet, and its only Terminal-Bench 4.0 value (vals.ai, 55.05 ± 1.82 at max) comes from a harness not comparable with GPT-6 Sol's Artificial Analysis one; the addendum's Broken Search Tool figure (2.08 % failure, https://deploymentsafety.openai.com/gpt-6-1-sol) is left out because a shipped honesty row would become the nearest honesty stand-in for unrelated sparse rungs; GPT-6 Sol at the same effort stands in for every dimension (the nearest stand-in, spec 1.2 §6.3)"
3060
- },
3061
- "gpt-6.1-sol#medium": {
3062
- "like": "gpt-6-sol#medium",
3063
- "note": "no DeepSWE 1.1 or Arena score is published for GPT-6.1 Sol yet, and its only Terminal-Bench 4.0 value (vals.ai, 55.05 ± 1.82 at max) comes from a harness not comparable with GPT-6 Sol's Artificial Analysis one; the addendum's Broken Search Tool figure (2.08 % failure, https://deploymentsafety.openai.com/gpt-6-1-sol) is left out because a shipped honesty row would become the nearest honesty stand-in for unrelated sparse rungs; GPT-6 Sol at the same effort stands in for every dimension (the nearest stand-in, spec 1.2 §6.3)"
3064
- },
3065
- "gpt-6.1-sol#high": {
3066
- "like": "gpt-6-sol#high",
3067
- "note": "no DeepSWE 1.1 or Arena score is published for GPT-6.1 Sol yet, and its only Terminal-Bench 4.0 value (vals.ai, 55.05 ± 1.82 at max) comes from a harness not comparable with GPT-6 Sol's Artificial Analysis one; the addendum's Broken Search Tool figure (2.08 % failure, https://deploymentsafety.openai.com/gpt-6-1-sol) is left out because a shipped honesty row would become the nearest honesty stand-in for unrelated sparse rungs; GPT-6 Sol at the same effort stands in for every dimension (the nearest stand-in, spec 1.2 §6.3)"
3068
- },
3069
- "gpt-6.1-sol#xhigh": {
3070
- "like": "gpt-6-sol#xhigh",
3071
- "note": "no DeepSWE 1.1 or Arena score is published for GPT-6.1 Sol yet, and its only Terminal-Bench 4.0 value (vals.ai, 55.05 ± 1.82 at max) comes from a harness not comparable with GPT-6 Sol's Artificial Analysis one; the addendum's Broken Search Tool figure (2.08 % failure, https://deploymentsafety.openai.com/gpt-6-1-sol) is left out because a shipped honesty row would become the nearest honesty stand-in for unrelated sparse rungs; GPT-6 Sol at the same effort stands in for every dimension (the nearest stand-in, spec 1.2 §6.3)"
3072
- },
3073
- "gpt-6.1-sol#max": {
3074
- "like": "gpt-6-sol#max",
3075
- "note": "no DeepSWE 1.1 or Arena score is published for GPT-6.1 Sol yet, and its only Terminal-Bench 4.0 value (vals.ai, 55.05 ± 1.82 at max) comes from a harness not comparable with GPT-6 Sol's Artificial Analysis one; the addendum's Broken Search Tool figure (2.08 % failure, https://deploymentsafety.openai.com/gpt-6-1-sol) is left out because a shipped honesty row would become the nearest honesty stand-in for unrelated sparse rungs; GPT-6 Sol at the same effort stands in for every dimension (the nearest stand-in, spec 1.2 §6.3)"
3076
- },
3077
- "gpt-6.1-sol#ultra": {
3078
- "like": "gpt-6-sol#ultra",
3079
- "note": "no DeepSWE 1.1 or Arena score is published for GPT-6.1 Sol yet, and its only Terminal-Bench 4.0 value (vals.ai, 55.05 ± 1.82 at max) comes from a harness not comparable with GPT-6 Sol's Artificial Analysis one; the addendum's Broken Search Tool figure (2.08 % failure, https://deploymentsafety.openai.com/gpt-6-1-sol) is left out because a shipped honesty row would become the nearest honesty stand-in for unrelated sparse rungs; GPT-6 Sol at the same effort stands in for every dimension (the nearest stand-in, spec 1.2 §6.3)"
3080
- }
3081
- },
3082
- "bars": {
3083
- "repo_code": {
3084
- "copy": {
3085
- "repo_code": 60.95
3086
- },
3087
- "build": {
3088
- "repo_code": 66.6
3089
- },
3090
- "logic": {
3091
- "repo_code": 67,
3092
- "honesty": 80.82,
3093
- "agentic": 0.08606
3094
- },
3095
- "hard": {
3096
- "repo_code": 70.9,
3097
- "honesty": 95.1,
3098
- "agentic": 0.1077
3099
- }
3100
- },
3101
- "terminal": {
3102
- "copy": {
3103
- "terminal": 40.15
3104
- },
3105
- "build": {
3106
- "terminal": 55.8
3107
- },
3108
- "logic": {
3109
- "terminal": 57.06,
3110
- "honesty": 80.82,
3111
- "agentic": 0.08606
3112
- },
3113
- "hard": {
3114
- "terminal": 61.35,
3115
- "honesty": 95.1,
3116
- "agentic": 0.1077
3117
- }
3118
- },
3119
- "ui": {
3120
- "copy": {
3121
- "frontend": 1539,
3122
- "repo_code": 60.95
3123
- },
3124
- "build": {
3125
- "frontend": 1617,
3126
- "repo_code": 66.6
3127
- },
3128
- "logic": {
3129
- "frontend": 1668,
3130
- "repo_code": 67,
3131
- "honesty": 80.82,
3132
- "agentic": 0.08606
3133
- },
3134
- "hard": {
3135
- "frontend": 1751,
3136
- "repo_code": 70.9,
3137
- "honesty": 95.1,
3138
- "agentic": 0.1077
3139
- }
3140
- },
3141
- "prose": {
3142
- "copy": {
3143
- "repo_code": 60.95
3144
- },
3145
- "build": {
3146
- "repo_code": 66.6
3147
- },
3148
- "logic": {
3149
- "repo_code": 67,
3150
- "honesty": 80.82
3151
- },
3152
- "hard": {
3153
- "repo_code": 70.9,
3154
- "honesty": 95.1
3155
- }
3156
- },
3157
- "research": {
3158
- "copy": {
3159
- "repo_code": 60.95
3160
- },
3161
- "build": {
3162
- "repo_code": 66.6
3163
- },
3164
- "logic": {
3165
- "repo_code": 67,
3166
- "honesty": 80.82
3167
- },
3168
- "hard": {
3169
- "repo_code": 70.9,
3170
- "honesty": 95.1
3171
- }
3172
- }
3173
- },
3174
- "barsWhy": {
3175
- "repo_code": {
3176
- "copy": "the 25th percentile of 11 rungs measured or better on DeepSWE 1.1 (2026-09-28): 60.95",
3177
- "build": "the median of 11 rungs measured or better on DeepSWE 1.1 (2026-09-28): 66.6",
3178
- "logic": "the 60th percentile of 11 rungs measured or better on DeepSWE 1.1 (2026-09-28): 67",
3179
- "hard": "the 75th percentile of 11 rungs measured or better on DeepSWE 1.1 (2026-09-28): 70.9"
3180
- },
3181
- "terminal": {
3182
- "copy": "the 25th percentile of 7 rungs measured or better on Terminal-Bench 4.0 (2026-09-28): 40.15",
3183
- "build": "the median of 7 rungs measured or better on Terminal-Bench 4.0 (2026-09-28): 55.8",
3184
- "logic": "the 60th percentile of 7 rungs measured or better on Terminal-Bench 4.0 (2026-09-28): 57.06",
3185
- "hard": "the 75th percentile of 7 rungs measured or better on Terminal-Bench 4.0 (2026-09-28): 61.35"
3186
- },
3187
- "honesty": {
3188
- "logic": "the 60th percentile of 5 rungs measured or better on OpenAI Broken Search Tool, 100 minus the failure rate GPT-6 system card appendix 11.6.4.2 (2026-09-28): 80.82",
3189
- "hard": "the 75th percentile of 5 rungs measured or better on OpenAI Broken Search Tool, 100 minus the failure rate GPT-6 system card appendix 11.6.4.2 (2026-09-28): 95.1"
3190
- },
3191
- "agentic": {
3192
- "logic": "the 60th percentile of 8 rungs measured or better on Arena agent, net improvement lmarena-ai/leaderboard-dataset (2026-09-28): 0.08606",
3193
- "hard": "the 75th percentile of 8 rungs measured or better on Arena agent, net improvement lmarena-ai/leaderboard-dataset (2026-09-28): 0.1077"
3194
- },
3195
- "frontend": {
3196
- "copy": "the 25th percentile of 9 rungs measured or better on Arena WebDev rating lmarena-ai/leaderboard-dataset (2026-09-28): 1539",
3197
- "build": "the median of 9 rungs measured or better on Arena WebDev rating lmarena-ai/leaderboard-dataset (2026-09-28): 1617",
3198
- "logic": "the 60th percentile of 9 rungs measured or better on Arena WebDev rating lmarena-ai/leaderboard-dataset (2026-09-28): 1668",
3199
- "hard": "the 75th percentile of 9 rungs measured or better on Arena WebDev rating lmarena-ai/leaderboard-dataset (2026-09-28): 1751"
3200
- }
3201
- }
3202
- }