omp-llm-role 1.0.0 → 1.1.0

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 (49) hide show
  1. package/README.md +179 -11
  2. package/agents/designer.md +4 -3
  3. package/docs/dev/README.md +44 -7
  4. package/docs/dev/agent-authoring.md +59 -1
  5. package/docs/dev/architecture.md +84 -27
  6. package/docs/dev/ci.md +152 -0
  7. package/docs/dev/current-state.md +327 -24
  8. package/docs/dev/data-sources.md +202 -38
  9. package/docs/dev/explorer.md +157 -46
  10. package/docs/dev/quirks.md +45 -4
  11. package/docs/dev/releasing.md +129 -17
  12. package/docs/dev/scoring.md +789 -28
  13. package/docs/dev/testing.md +206 -0
  14. package/docs/dev/writes-and-state.md +176 -28
  15. package/docs/llm-role-rankings.md +141 -140
  16. package/package.json +754 -80
  17. package/skills/omp-llm-role-create-agent/SKILL.md +7 -1
  18. package/src/agent-architect.ts +148 -19
  19. package/src/agent-create.ts +800 -40
  20. package/src/agent-file.ts +2 -5
  21. package/src/architect-provenance.ts +16 -0
  22. package/src/availability.ts +104 -6
  23. package/src/benchmark-author.ts +9 -4
  24. package/src/benchmark-sources.ts +283 -18
  25. package/src/cli/create-role.ts +29 -1
  26. package/src/cli/llm-role-rank.ts +40 -1
  27. package/src/config-edit.ts +207 -80
  28. package/src/engine.ts +409 -46
  29. package/src/error-report.ts +562 -0
  30. package/src/explorer/boot.ts +36 -24
  31. package/src/explorer/explain.ts +110 -13
  32. package/src/explorer/scopes.ts +110 -0
  33. package/src/explorer/server.ts +112 -32
  34. package/src/extension.ts +407 -34
  35. package/src/features.ts +103 -0
  36. package/src/project-registry.ts +65 -0
  37. package/src/project-setup.ts +382 -0
  38. package/src/prompts/agent-creation-architect.md +2 -0
  39. package/src/prompts/project-profile-architect.md +33 -0
  40. package/src/prompts/project-profile-user.md +8 -0
  41. package/src/role-archetypes.ts +12 -2
  42. package/src/role-settings.ts +22 -0
  43. package/src/settings.ts +205 -24
  44. package/src/state.ts +1 -1
  45. package/src/updater.ts +150 -31
  46. package/web/app.js +411 -12
  47. package/web/index.html +1 -0
  48. package/web/style.css +44 -1
  49. package/docs/dev/spec.md +0 -633
package/docs/llm-role-rankings.md CHANGED
@@ -1,9 +1,9 @@
1
- llm-stats.com best-fit ranking per omp model role — 400 models, 2026-10-02
1
+ llm-stats.com best-fit ranking per omp model role — 407 models, 2026-10-08
2
2
  Value ranking per role: each metric is cardinal-normalized with fixed anchors
3
3
  (no ranks): llm-stats index_* affine (v+20)/80 (interval scale, observed −16..+60),
4
4
  benchmarks chance-anchored pass rates, throughput log-anchored 10..300 tok/s.
5
5
  q = Σ weight × metric over the quality metrics (weights renormalized excluding
6
- price); value = q − λ·$/M sorts each role. λ = price-weight share ÷ $20, per-role
6
+ price); value = q − λ·$/M sorts each role. λ = w/(1−w) ÷ $20 (w = price weight), per-role
7
7
  override via plugin settings roles.<role>.lambda. Metric columns show weighted
8
8
  contributions and sum to q; — = metric missing (contributes 0).
9
9
  Roles with a thinking level rank on the thinking-adjusted price: the billed
@@ -21,178 +21,179 @@ agon=Design Arena `agents/agon_webapps` Elo from the same endpoint: context
21
21
  only, unweighted (22/87 of the designer pool, below the ~35–40% coverage
22
22
  bar at which a metric earns weight).
23
23
  ★ = Pareto-frontier: no eligible model is both cheaper and better (q).
24
- Throughput + price: OpenRouter per-provider routes under the default price-based routing (weight 1/price² over stable standard-tier providers; p50 tok/s of routed traffic; $/M 3:1 in:out), throughput 152/400, priced 151; models without OpenRouter throughput or a billed route are not ranked.
24
+ Throughput + price: OpenRouter per-provider routes under the default price-based routing (weight 1/price² over stable standard-tier providers; p50 tok/s of routed traffic; $/M 3:1 in:out), throughput 155/407, priced 153; models without OpenRouter throughput or a billed route are not ranked.
25
25
 
26
26
  ## @default — Main workhorse: strong general coding-agent quality, sane cost
27
- eligible: 143 — λ 0.00556 $/quality-point (thinking ×1.857)
28
-
29
- | # | value | q | model | org | $/M | tok/s | ctx | gen 35.8% | rea 18.9% | code 18.9% | ag 12.6% | tool 10.5% | tput 3.2% |
30
- | -: | ----: | ----: | --------------------- | ------------------------- | ----: | ----: | ---: | --------: | --------: | ---------: | -------: | ---------: | --------: |
31
- | 1 | 0.810 | 0.812 | ★ DeepSeek-V4.1-Flash | DeepSeek | $0.45 | 87 | 1.0M | 0.318 | 0.163 | 0.149 | 0.094 | 0.068 | 0.020 |
32
- | 2 | 0.803 | 0.824 | ★ Muse Spark 1.3 | Meta | $3.71 | 81 | 1.0M | 0.330 | 0.169 | 0.145 | 0.094 | 0.066 | 0.019 |
33
- | 3 | 0.803 | 0.811 | GLM-5.3 | Zhipu AI | $1.50 | 59 | 1.0M | 0.322 | 0.169 | 0.145 | 0.092 | 0.066 | 0.016 |
34
- | 4 | 0.782 | 0.788 | DeepSeek-V4-Pro-0813 | DeepSeek | $0.94 | 49 | 1.0M | 0.318 | 0.164 | 0.140 | 0.089 | 0.063 | 0.015 |
35
- | 5 | 0.775 | 0.780 | MiMo-V2.6-Pro | Xiaomi | $0.99 | 26 | 1.0M | 0.313 | 0.158 | 0.144 | 0.091 | 0.065 | 0.009 |
36
- | 6 | 0.771 | 0.834 | ★ GPT-5.6 Sol | OpenAI | $11.4 | 60 | 1.1M | 0.331 | 0.173 | 0.154 | 0.095 | 0.064 | 0.017 |
37
- | 7 | 0.762 | 0.764 | ★ GLM-5.3-Flash | Zhipu AI | $0.36 | 41 | 1.0M | 0.311 | 0.162 | 0.126 | 0.089 | 0.064 | 0.013 |
38
- | 8 | 0.761 | 0.764 | Qwen3.8 Flash | Alibaba Cloud / Qwen Team | $0.43 | 46 | 1.0M | 0.307 | 0.162 | 0.130 | 0.085 | 0.064 | 0.014 |
39
- | 9 | 0.761 | 0.774 | Hy4 preview | Tencent | $2.32 | 15 | — | 0.315 | 0.166 | 0.138 | 0.088 | 0.063 | 0.004 |
40
- | 10 | 0.757 | 0.778 | Muse Spark 1.1 | Meta | $3.71 | 170 | 1.0M | 0.309 | 0.165 | 0.127 | 0.085 | 0.065 | 0.026 |
27
+ eligible: 146 — λ 0.00556 $/quality-point (thinking ×1.857)
28
+
29
+ | # | value | q | model | org | $/M | tok/s | ctx | gen 54.7% | code 18.9% | ag 12.6% | tool 10.5% | tput 3.2% |
30
+ | -: | ----: | ----: | --------------------- | ----------- | ----: | ----: | ---: | --------: | ---------: | -------: | ---------: | --------: |
31
+ | 1 | 0.808 | 0.810 | ★ DeepSeek-V4.1-Flash | DeepSeek | $0.42 | 57 | 1.0M | 0.485 | 0.148 | 0.094 | 0.068 | 0.016 |
32
+ | 2 | 0.801 | 0.822 | ★ Muse Spark 1.3 | Meta | $3.71 | 63 | 1.0M | 0.501 | 0.145 | 0.093 | 0.066 | 0.017 |
33
+ | 3 | 0.794 | 0.813 | ★ GLM-5.3 | Zhipu AI | $3.37 | 89 | 1.0M | 0.490 | 0.144 | 0.091 | 0.066 | 0.020 |
34
+ | 4 | 0.782 | 0.787 | MiMo-V2.6-Pro | Xiaomi | $1.01 | 30 | 1.0M | 0.479 | 0.143 | 0.090 | 0.065 | 0.010 |
35
+ | 5 | 0.777 | 0.793 | DeepSeek-V4-Pro-0813 | DeepSeek | $2.79 | 85 | 1.0M | 0.483 | 0.139 | 0.088 | 0.062 | 0.020 |
36
+ | 6 | 0.774 | 0.818 | GPT-6.1 Sol | OpenAI | $7.84 | 39 | 1.1M | 0.508 | 0.146 | 0.094 | 0.057 | 0.013 |
37
+ | 7 | 0.772 | 0.814 | Kimi K3 | Moonshot AI | $7.52 | 58 | 1.0M | 0.494 | 0.146 | 0.091 | 0.067 | 0.016 |
38
+ | 8 | 0.767 | 0.831 | ★ GPT-5.6 Sol | OpenAI | $11.4 | 61 | 1.1M | 0.503 | 0.152 | 0.094 | 0.064 | 0.017 |
39
+ | 9 | 0.763 | 0.776 | Hy4 preview | Tencent | $2.32 | 33 | — | 0.478 | 0.137 | 0.087 | 0.062 | 0.011 |
40
+ | 10 | 0.759 | 0.762 | ★ GLM-5.3-Flash | Zhipu AI | $0.41 | 37 | 1.0M | 0.472 | 0.126 | 0.088 | 0.064 | 0.012 |
41
41
 
42
42
  ## @smol — Fast lightweight model: cheap and quick, still competent
43
- eligible: 143 — λ 0.02143 $/quality-point
44
-
45
- | # | value | q | model | org | $/M | tok/s | ctx | tput 40% | gen 28.6% | code 17.1% | tool 14.3% |
46
- | -: | ----: | ----: | ---------------------------- | ----------- | ----: | ----: | ---: | -------: | --------: | ---------: | ---------: |
47
- | 1 | 0.740 | 0.783 | ★ Muse Spark 1.1 | Meta | $2.00 | 170 | 1.0M | 0.333 | 0.247 | 0.115 | 0.088 |
48
- | 2 | 0.730 | 0.735 | ★ DeepSeek-V4.1-Flash | DeepSeek | $0.24 | 87 | 1.0M | 0.254 | 0.254 | 0.135 | 0.092 |
49
- | 3 | 0.688 | 0.731 | Muse Spark 1.3 | Meta | $2.00 | 81 | 1.0M | 0.246 | 0.263 | 0.131 | 0.090 |
50
- | 4 | 0.685 | 0.717 | Gemini 3.8 Flash | Google | $1.50 | 99 | 1.0M | 0.270 | 0.248 | 0.123 | 0.077 |
51
- | 5 | 0.669 | 0.679 | DeepSeek-V4-Flash-Vision-Exp | DeepSeek | $0.46 | 89 | 1.0M | 0.257 | 0.232 | 0.116 | 0.074 |
52
- | 6 | 0.669 | 0.686 | GLM-5.3 | Zhipu AI | $0.81 | 59 | 1.0M | 0.208 | 0.257 | 0.132 | 0.090 |
53
- | 7 | 0.669 | 0.701 | Gemini 3.7 Flash | Google | $1.50 | 99 | 1.0M | 0.269 | 0.242 | 0.115 | 0.075 |
54
- | 8 | 0.654 | 0.728 | Gemini 3.5 Flash | Google | $3.47 | 138 | 1.0M | 0.309 | 0.223 | 0.108 | 0.088 |
55
- | 9 | 0.648 | 0.649 | ★ Ling 3.0 Flash | InclusionAI | $0.04 | 115 | 131k | 0.287 | 0.201 | 0.096 | 0.064 |
56
- | 10 | 0.642 | 0.652 | DeepSeek-V4-Pro-0813 | DeepSeek | $0.51 | 49 | 1.0M | 0.187 | 0.254 | 0.127 | 0.085 |
43
+ eligible: 146 — λ 0.04091 $/quality-point
44
+
45
+ | # | value | q | model | org | $/M | tok/s | ctx | tput 40% | gen 28.6% | code 17.1% | tool 14.3% |
46
+ | -: | ----: | ----: | ---------------------------- | ------------------------- | ----: | ----: | ---: | -------: | --------: | ---------: | ---------: |
47
+ | 1 | 0.687 | 0.769 | ★ Muse Spark 1.1 | Meta | $2.00 | 154 | 1.0M | 0.322 | 0.245 | 0.114 | 0.088 |
48
+ | 2 | 0.675 | 0.684 | ★ DeepSeek-V4.1-Flash | DeepSeek | $0.22 | 57 | 1.0M | 0.205 | 0.253 | 0.134 | 0.092 |
49
+ | 3 | 0.659 | 0.733 | ★ GLM-5.3 | Zhipu AI | $1.81 | 89 | 1.0M | 0.256 | 0.256 | 0.131 | 0.090 |
50
+ | 4 | 0.656 | 0.675 | DeepSeek-V4-Flash-Vision-Exp | DeepSeek | $0.46 | 87 | 1.0M | 0.255 | 0.231 | 0.115 | 0.074 |
51
+ | 5 | 0.654 | 0.655 | ★ Ling 3.0 Flash | InclusionAI | $0.04 | 124 | 131k | 0.296 | 0.200 | 0.096 | 0.064 |
52
+ | 6 | 0.654 | 0.715 | ★ DeepSeek-V4-Pro-0813 | DeepSeek | $1.50 | 85 | 1.0M | 0.252 | 0.252 | 0.126 | 0.085 |
53
+ | 7 | 0.653 | 0.673 | GPT-5.6 Luna | OpenAI | $0.47 | 83 | 1.1M | 0.249 | 0.230 | 0.118 | 0.076 |
54
+ | 8 | 0.647 | 0.662 | Mercury 2 | Inception | $0.38 | 302 | 128k | 0.400 | 0.155 | 0.069 | 0.038 |
55
+ | 9 | 0.643 | 0.652 | Qwen3.8 Flash | Alibaba Cloud / Qwen Team | $0.23 | 57 | 1.0M | 0.205 | 0.244 | 0.117 | 0.087 |
56
+ | 10 | 0.636 | 0.645 | GPT-6 Luna | OpenAI | $0.21 | 73 | 1.1M | 0.234 | 0.228 | 0.108 | 0.074 |
57
57
 
58
58
  ## @slow — Most capable model for hard problems; cost and speed as tiebreakers
59
- eligible: 143 — λ 0.00263 $/quality-point (thinking ×2.714)
60
-
61
- | # | value | q | model | org | $/M | tok/s | ctx | gen 27.4% | rea 27.4% | code 18.9% | ag 13.7% | math 8.4% | tput 4.2% |
62
- | -: | ----: | ----: | --------------------- | ------------------------- | ----: | ----: | ---: | --------: | --------: | ---------: | -------: | --------: | --------: |
63
- | 1 | 0.816 | 0.822 | ★ GLM-5.3 | Zhipu AI | $2.19 | 59 | 1.0M | 0.246 | 0.244 | 0.145 | 0.100 | 0.064 | 0.022 |
64
- | 2 | 0.812 | 0.814 | ★ DeepSeek-V4.1-Flash | DeepSeek | $0.66 | 87 | 1.0M | 0.244 | 0.235 | 0.149 | 0.102 | 0.058 | 0.027 |
65
- | 3 | 0.796 | 0.840 | ★ GPT-5.6 Sol | OpenAI | $16.6 | 60 | 1.1M | 0.253 | 0.250 | 0.154 | 0.103 | 0.058 | 0.022 |
66
- | 4 | 0.794 | 0.798 | DeepSeek-V4-Pro-0813 | DeepSeek | $1.38 | 49 | 1.0M | 0.243 | 0.236 | 0.140 | 0.096 | 0.063 | 0.020 |
67
- | 5 | 0.782 | 0.821 | Kimi K3 | Moonshot AI | $14.7 | 47 | 1.0M | 0.248 | 0.244 | 0.147 | 0.099 | 0.064 | 0.019 |
68
- | 6 | 0.779 | 0.793 | Muse Spark 1.1 | Meta | $5.43 | 170 | 1.0M | 0.237 | 0.238 | 0.127 | 0.092 | 0.064 | 0.035 |
69
- | 7 | 0.774 | 0.783 | Hy4 preview | Tencent | $3.39 | 15 | — | 0.241 | 0.240 | 0.138 | 0.095 | 0.063 | 0.005 |
70
- | 8 | 0.769 | 0.770 | ★ GLM-5.3-Flash | Zhipu AI | $0.53 | 41 | 1.0M | 0.238 | 0.234 | 0.126 | 0.096 | 0.059 | 0.017 |
71
- | 9 | 0.767 | 0.842 | ★ Claude Opus 5 | Anthropic | $28.4 | 61 | 1.0M | 0.255 | 0.253 | 0.144 | 0.101 | 0.066 | 0.022 |
72
- | 10 | 0.765 | 0.766 | Qwen3.8 Flash | Alibaba Cloud / Qwen Team | $0.62 | 46 | 1.0M | 0.235 | 0.234 | 0.130 | 0.092 | 0.055 | 0.019 |
59
+ eligible: 146 — λ 0.00263 $/quality-point (thinking ×2.714)
60
+
61
+ | # | value | q | model | org | $/M | tok/s | ctx | gen 63.2% | code 18.9% | ag 13.7% | tput 4.2% |
62
+ | -: | ----: | ----: | --------------------- | ----------- | ----: | ----: | ---: | --------: | ---------: | -------: | --------: |
63
+ | 1 | 0.832 | 0.846 | ★ Muse Spark 1.3 | Meta | $5.43 | 63 | 1.0M | 0.578 | 0.145 | 0.101 | 0.023 |
64
+ | 2 | 0.829 | 0.830 | ★ DeepSeek-V4.1-Flash | DeepSeek | $0.61 | 57 | 1.0M | 0.559 | 0.148 | 0.101 | 0.022 |
65
+ | 3 | 0.823 | 0.836 | ★ GLM-5.3 | Zhipu AI | $4.92 | 89 | 1.0M | 0.566 | 0.144 | 0.099 | 0.027 |
66
+ | 4 | 0.821 | 0.851 | ★ GPT-6.1 Sol | OpenAI | $11.5 | 39 | 1.1M | 0.586 | 0.146 | 0.101 | 0.017 |
67
+ | 5 | 0.814 | 0.857 | ★ GPT-5.6 Sol | OpenAI | $16.6 | 61 | 1.1M | 0.581 | 0.152 | 0.102 | 0.022 |
68
+ | 6 | 0.808 | 0.819 | DeepSeek-V4-Pro-0813 | DeepSeek | $4.08 | 85 | 1.0M | 0.558 | 0.139 | 0.095 | 0.027 |
69
+ | 7 | 0.807 | 0.836 | Kimi K3 | Moonshot AI | $11 | 58 | 1.0M | 0.570 | 0.146 | 0.099 | 0.022 |
70
+ | 8 | 0.803 | 0.806 | MiMo-V2.6-Pro | Xiaomi | $1.47 | 30 | 1.0M | 0.552 | 0.143 | 0.098 | 0.013 |
71
+ | 9 | 0.789 | 0.798 | Hy4 preview | Tencent | $3.39 | 33 | — | 0.552 | 0.137 | 0.094 | 0.015 |
72
+ | 10 | 0.782 | 0.793 | Gemini 3.8 Flash | Google | $4.07 | 78 | 1.0M | 0.543 | 0.134 | 0.091 | 0.025 |
73
73
 
74
74
  ## @vision — Image understanding: vision index dominates
75
75
  eligible: 75 — λ 0.00682 $/quality-point (thinking ×1.857)
76
76
 
77
- | # | value | q | model | org | $/M | tok/s | ctx | vis 54.2% | gen 20.8% | rea 15.6% | code 6.3% | tput 3.1% |
78
- | -: | ----: | ----: | ------------------- | ------------------------- | ----: | ----: | ---: | --------: | --------: | --------: | --------: | --------: |
79
- | 1 | 0.731 | 0.734 | ★ Qwen3.8 Flash | Alibaba Cloud / Qwen Team | $0.43 | 46 | 1.0M | 0.364 | 0.179 | 0.134 | 0.043 | 0.014 |
80
- | 2 | 0.730 | 0.756 | ★ Muse Spark 1.1 | Meta | $3.71 | 170 | 1.0M | 0.371 | 0.180 | 0.136 | 0.042 | 0.026 |
81
- | 3 | 0.729 | 0.732 | DeepSeek-V4.1-Flash | DeepSeek | $0.45 | 87 | 1.0M | 0.343 | 0.185 | 0.134 | 0.049 | 0.020 |
82
- | 4 | 0.720 | 0.739 | ★ Gemini 3.8 Flash | Google | $2.79 | 99 | 1.0M | 0.363 | 0.181 | 0.128 | 0.045 | 0.021 |
83
- | 5 | 0.713 | 0.715 | ★ GLM-5.3-Flash | Zhipu AI | $0.36 | 41 | 1.0M | 0.346 | 0.181 | 0.133 | 0.042 | 0.013 |
84
- | 6 | 0.702 | 0.721 | Gemini 3.7 Flash | Google | $2.79 | 99 | 1.0M | 0.354 | 0.176 | 0.128 | 0.042 | 0.021 |
85
- | 7 | 0.701 | 0.770 | ★ Kimi K3 | Moonshot AI | $10.1 | 47 | 1.0M | 0.380 | 0.189 | 0.139 | 0.048 | 0.014 |
86
- | 8 | 0.697 | 0.706 | Qwen3.8-27B | Alibaba Cloud / Qwen Team | $1.38 | 47 | 262k | 0.359 | 0.169 | 0.126 | 0.040 | 0.014 |
87
- | 9 | 0.696 | 0.774 | ★ GPT-5.6 Sol | OpenAI | $11.4 | 60 | 1.1M | 0.371 | 0.193 | 0.143 | 0.051 | 0.016 |
88
- | 10 | 0.688 | 0.713 | Muse Spark 1.3 | Meta | $3.71 | 81 | 1.0M | 0.314 | 0.192 | 0.139 | 0.048 | 0.019 |
89
-
90
- ## @plan — Planning: reasoning, math, long-context coherence
91
- eligible: 143 — λ 0.00682 $/quality-point (thinking ×1.857)
92
-
93
- | # | value | q | model | org | $/M | tok/s | ctx | rea 35.6% | gen 28.9% | lc 15.6% | math 15.6% | tput 4.4% |
94
- | -: | ----: | ----: | -------------------- | ------------------------- | ----: | ----: | ---: | --------: | --------: | -------: | ---------: | --------: |
95
- | 1 | 0.751 | 0.754 | ★ Qwen3.8 Flash | Alibaba Cloud / Qwen Team | $0.43 | 46 | 1.0M | 0.305 | 0.248 | 0.079 | 0.102 | 0.020 |
96
- | 2 | 0.747 | 0.763 | ★ Hy4 preview | Tencent | $2.32 | 15 | — | 0.312 | 0.254 | 0.076 | 0.116 | 0.005 |
97
- | 3 | 0.744 | 0.822 | ★ GPT-5.6 Sol | OpenAI | $11.4 | 60 | 1.1M | 0.324 | 0.267 | 0.099 | 0.107 | 0.023 |
98
- | 4 | 0.739 | 0.749 | GLM-5.3 | Zhipu AI | $1.50 | 59 | 1.0M | 0.317 | 0.260 | 0.030 | 0.119 | 0.023 |
99
- | 5 | 0.725 | 0.728 | DeepSeek-V4.1-Flash | DeepSeek | $0.45 | 87 | 1.0M | 0.305 | 0.257 | 0.030 | 0.107 | 0.028 |
100
- | 6 | 0.724 | 0.730 | DeepSeek-V4-Pro-0813 | DeepSeek | $0.94 | 49 | 1.0M | 0.307 | 0.256 | 0.030 | 0.116 | 0.021 |
101
- | 7 | 0.723 | 0.726 | Hy3 | Tencent | $0.47 | 61 | 262k | 0.279 | 0.227 | 0.088 | 0.109 | 0.024 |
102
- | 8 | 0.719 | 0.744 | Muse Spark 1.1 | Meta | $3.71 | 170 | 1.0M | 0.310 | 0.250 | 0.030 | 0.118 | 0.037 |
103
- | 9 | 0.713 | 0.774 | ★ GPT-5.6 Terra | OpenAI | $8.82 | 73 | 1.1M | 0.303 | 0.253 | 0.089 | 0.102 | 0.026 |
104
- | 10 | 0.710 | 0.713 | ★ GLM-5.3-Flash | Zhipu AI | $0.36 | 41 | 1.0M | 0.304 | 0.251 | 0.030 | 0.110 | 0.018 |
77
+ | # | value | q | model | org | $/M | tok/s | ctx | vis 54.2% | gen 20.8% | rea 15.6% | code 6.3% | tput 3.1% |
78
+ | -: | ----: | ----: | --------------------- | ------------------------- | ----: | ----: | ---: | --------: | --------: | --------: | --------: | --------: |
79
+ | 1 | 0.731 | 0.734 | ★ Qwen3.8 Flash | Alibaba Cloud / Qwen Team | $0.43 | 57 | 1.0M | 0.364 | 0.178 | 0.133 | 0.042 | 0.016 |
80
+ | 2 | 0.727 | 0.752 | ★ Muse Spark 1.1 | Meta | $3.71 | 154 | 1.0M | 0.371 | 0.179 | 0.135 | 0.042 | 0.025 |
81
+ | 3 | 0.727 | 0.729 | ★ DeepSeek-V4.1-Flash | DeepSeek | $0.42 | 57 | 1.0M | 0.347 | 0.184 | 0.134 | 0.049 | 0.016 |
82
+ | 4 | 0.721 | 0.774 | ★ GPT-6.1 Sol | OpenAI | $7.84 | 39 | 1.1M | 0.379 | 0.193 | 0.141 | 0.048 | 0.012 |
83
+ | 5 | 0.719 | 0.770 | ★ Kimi K3 | Moonshot AI | $7.52 | 58 | 1.0M | 0.379 | 0.188 | 0.139 | 0.048 | 0.016 |
84
+ | 6 | 0.715 | 0.734 | ★ Gemini 3.8 Flash | Google | $2.79 | 78 | 1.0M | 0.365 | 0.179 | 0.128 | 0.044 | 0.019 |
85
+ | 7 | 0.708 | 0.710 | ★ GLM-5.3-Flash | Zhipu AI | $0.41 | 37 | 1.0M | 0.344 | 0.180 | 0.133 | 0.041 | 0.012 |
86
+ | 8 | 0.701 | 0.720 | Gemini 3.7 Flash | Google | $2.79 | 91 | 1.0M | 0.355 | 0.175 | 0.128 | 0.042 | 0.020 |
87
+ | 9 | 0.697 | 0.707 | Qwen3.8-27B | Alibaba Cloud / Qwen Team | $1.48 | 52 | 262k | 0.359 | 0.168 | 0.125 | 0.039 | 0.015 |
88
+ | 10 | 0.693 | 0.771 | GPT-5.6 Sol | OpenAI | $11.4 | 61 | 1.1M | 0.370 | 0.192 | 0.142 | 0.050 | 0.017 |
89
+
90
+ ## @plan — Planning: reasoning and long-context coherence, priced so the cheapest capable planner wins
91
+ eligible: 146 — λ 0.01667 $/quality-point (thinking ×1.857)
92
+
93
+ | # | value | q | model | org | $/M | tok/s | ctx | gen 80% | lc 15.6% | tput 4.4% |
94
+ | -: | ----: | ----: | --------------------- | ------------------------- | ----: | ----: | ---: | ------: | -------: | --------: |
95
+ | 1 | 0.795 | 0.857 | ★ Muse Spark 1.3 | Meta | $3.71 | 63 | 1.0M | 0.732 | 0.101 | 0.024 |
96
+ | 2 | 0.779 | 0.786 | ★ Qwen3.8 Flash | Alibaba Cloud / Qwen Team | $0.43 | 57 | 1.0M | 0.683 | 0.079 | 0.023 |
97
+ | 3 | 0.761 | 0.807 | ★ Gemini 3.8 Flash | Google | $2.79 | 78 | 1.0M | 0.687 | 0.093 | 0.027 |
98
+ | 4 | 0.755 | 0.762 | ★ DeepSeek-V4.1-Flash | DeepSeek | $0.42 | 57 | 1.0M | 0.709 | 0.030 | 0.023 |
99
+ | 5 | 0.754 | 0.801 | Gemini 3.7 Flash | Google | $2.79 | 91 | 1.0M | 0.672 | 0.100 | 0.029 |
100
+ | 6 | 0.751 | 0.790 | ★ Hy4 preview | Tencent | $2.32 | 33 | — | 0.699 | 0.076 | 0.015 |
101
+ | 7 | 0.737 | 0.751 | GPT-5.6 Luna | OpenAI | $0.88 | 83 | 1.1M | 0.644 | 0.080 | 0.028 |
102
+ | 8 | 0.731 | 0.738 | ★ GLM-5.3-Flash | Zhipu AI | $0.41 | 37 | 1.0M | 0.690 | 0.030 | 0.017 |
103
+ | 9 | 0.727 | 0.744 | MiMo-V2.6-Pro | Xiaomi | $1.01 | 30 | 1.0M | 0.699 | 0.030 | 0.014 |
104
+ | 10 | 0.725 | 0.733 | Hy3 | Tencent | $0.47 | 52 | 262k | 0.626 | 0.086 | 0.021 |
105
105
 
106
106
  ## @commit — Commit messages: cheap and fast with decent general quality
107
- eligible: 143 — λ 0.02692 $/quality-point
108
-
109
- | # | value | q | model | org | $/M | tok/s | ctx | tput 41.5% | gen 43.1% | code 15.4% |
110
- | -: | ----: | ----: | ---------------------------- | ----------- | ----: | ----: | ---: | ---------: | --------: | ---------: |
111
- | 1 | 0.768 | 0.822 | ★ Muse Spark 1.1 | Meta | $2.00 | 170 | 1.0M | 0.346 | 0.372 | 0.103 |
112
- | 2 | 0.762 | 0.768 | ★ DeepSeek-V4.1-Flash | DeepSeek | $0.24 | 87 | 1.0M | 0.264 | 0.383 | 0.121 |
113
- | 3 | 0.724 | 0.765 | Gemini 3.8 Flash | Google | $1.50 | 99 | 1.0M | 0.280 | 0.375 | 0.110 |
114
- | 4 | 0.717 | 0.770 | Muse Spark 1.3 | Meta | $2.00 | 81 | 1.0M | 0.255 | 0.397 | 0.118 |
115
- | 5 | 0.709 | 0.721 | DeepSeek-V4-Flash-Vision-Exp | DeepSeek | $0.46 | 89 | 1.0M | 0.267 | 0.350 | 0.104 |
116
- | 6 | 0.706 | 0.746 | Gemini 3.7 Flash | Google | $1.50 | 99 | 1.0M | 0.279 | 0.364 | 0.103 |
117
- | 7 | 0.700 | 0.722 | GLM-5.3 | Zhipu AI | $0.81 | 59 | 1.0M | 0.216 | 0.388 | 0.118 |
118
- | 8 | 0.686 | 0.687 | ★ Ling 3.0 Flash | InclusionAI | $0.04 | 115 | 131k | 0.298 | 0.303 | 0.086 |
119
- | 9 | 0.677 | 0.690 | DeepSeek-V4-Pro-0813 | DeepSeek | $0.51 | 49 | 1.0M | 0.194 | 0.382 | 0.114 |
120
- | 10 | 0.676 | 0.681 | GPT-6 Luna | OpenAI | $0.21 | 69 | 1.1M | 0.237 | 0.346 | 0.098 |
107
+ eligible: 146 — λ 0.04091 $/quality-point
108
+
109
+ | # | value | q | model | org | $/M | tok/s | ctx | tput 41.5% | gen 43.1% | code 15.4% |
110
+ | -: | ----: | ----: | ------------------------------ | ----------- | ----: | ----: | ---: | ---------: | --------: | ---------: |
111
+ | 1 | 0.725 | 0.807 | ★ Muse Spark 1.1 | Meta | $2.00 | 154 | 1.0M | 0.334 | 0.370 | 0.103 |
112
+ | 2 | 0.705 | 0.715 | ★ DeepSeek-V4.1-Flash | DeepSeek | $0.22 | 57 | 1.0M | 0.213 | 0.382 | 0.120 |
113
+ | 3 | 0.697 | 0.716 | ★ DeepSeek-V4-Flash-Vision-Exp | DeepSeek | $0.46 | 87 | 1.0M | 0.264 | 0.349 | 0.104 |
114
+ | 4 | 0.696 | 0.711 | Mercury 2 | Inception | $0.38 | 302 | 128k | 0.415 | 0.234 | 0.062 |
115
+ | 5 | 0.695 | 0.770 | ★ GLM-5.3 | Zhipu AI | $1.81 | 89 | 1.0M | 0.266 | 0.386 | 0.117 |
116
+ | 6 | 0.694 | 0.755 | ★ DeepSeek-V4-Pro-0813 | DeepSeek | $1.50 | 85 | 1.0M | 0.262 | 0.380 | 0.113 |
117
+ | 7 | 0.693 | 0.695 | ★ Ling 3.0 Flash | InclusionAI | $0.04 | 124 | 131k | 0.307 | 0.302 | 0.086 |
118
+ | 8 | 0.692 | 0.711 | GPT-5.6 Luna | OpenAI | $0.47 | 83 | 1.1M | 0.259 | 0.347 | 0.106 |
119
+ | 9 | 0.680 | 0.684 | Laguna S 2.1 | Poolside | $0.11 | 89 | 1.0M | 0.267 | 0.322 | 0.095 |
120
+ | 10 | 0.676 | 0.685 | GPT-6 Luna | OpenAI | $0.21 | 73 | 1.1M | 0.243 | 0.344 | 0.097 |
121
121
 
122
122
  ## @tiny — Background tasks (titles, memory): cheapest and fastest wins
123
- eligible: 143 — λ 0.03333 $/quality-point
123
+ eligible: 146 — λ 0.03333 $/quality-point
124
124
 
125
125
  | # | value | q | model | org | $/M | tok/s | ctx | tput 58.3% | gen 41.7% |
126
126
  | -: | ----: | ----: | ---------------------------- | ------------------------- | ----: | ----: | ---: | ---------: | --------: |
127
- | 1 | 0.779 | 0.846 | ★ Muse Spark 1.1 | Meta | $2.00 | 170 | 1.0M | 0.485 | 0.360 |
128
- | 2 | 0.735 | 0.737 | ★ Ling 3.0 Flash Fin | InclusionAI | $0.07 | 110 | 262k | 0.411 | 0.327 |
129
- | 3 | 0.734 | 0.742 | ★ DeepSeek-V4.1-Flash | DeepSeek | $0.24 | 87 | 1.0M | 0.371 | 0.371 |
130
- | 4 | 0.710 | 0.712 | ★ Ling 3.0 Flash | InclusionAI | $0.04 | 115 | 131k | 0.419 | 0.293 |
131
- | 5 | 0.705 | 0.755 | ★ Gemini 3.8 Flash | Google | $1.50 | 99 | 1.0M | 0.393 | 0.362 |
132
- | 6 | 0.698 | 0.714 | DeepSeek-V4-Flash-Vision-Exp | DeepSeek | $0.46 | 89 | 1.0M | 0.375 | 0.339 |
133
- | 7 | 0.695 | 0.745 | Gemini 3.7 Flash | Google | $1.50 | 99 | 1.0M | 0.392 | 0.352 |
134
- | 8 | 0.676 | 0.682 | Seed 2.0 Mini | ByteDance | $0.17 | 125 | 256k | 0.433 | 0.249 |
135
- | 9 | 0.676 | 0.743 | Muse Spark 1.3 | Meta | $2.00 | 81 | 1.0M | 0.359 | 0.384 |
136
- | 10 | 0.668 | 0.682 | Qwen3-Next-80B-A3B-Thinking | Alibaba Cloud / Qwen Team | $0.41 | 141 | — | 0.454 | 0.228 |
127
+ | 1 | 0.797 | 0.810 | ★ Mercury 2 | Inception | $0.38 | 302 | 128k | 0.583 | 0.226 |
128
+ | 2 | 0.765 | 0.767 | ★ Ling 3.0 Flash Fin | InclusionAI | $0.07 | 131 | 262k | 0.441 | 0.326 |
129
+ | 3 | 0.760 | 0.827 | ★ Muse Spark 1.1 | Meta | $2.00 | 154 | 1.0M | 0.469 | 0.358 |
130
+ | 4 | 0.722 | 0.723 | ★ Ling 3.0 Flash | InclusionAI | $0.04 | 124 | 131k | 0.431 | 0.292 |
131
+ | 5 | 0.697 | 0.716 | Gemini 3.1 Flash-Lite | Google | $0.56 | 139 | 1.0M | 0.451 | 0.264 |
132
+ | 6 | 0.695 | 0.745 | Gemini 3.6 Flash | Google | $1.50 | 115 | 1.0M | 0.419 | 0.326 |
133
+ | 7 | 0.693 | 0.708 | DeepSeek-V4-Flash-Vision-Exp | DeepSeek | $0.46 | 87 | 1.0M | 0.371 | 0.337 |
134
+ | 8 | 0.687 | 0.701 | Qwen3-Next-80B-A3B-Thinking | Alibaba Cloud / Qwen Team | $0.41 | 158 | — | 0.473 | 0.228 |
135
+ | 9 | 0.687 | 0.747 | GLM-5.3 | Zhipu AI | $1.81 | 89 | 1.0M | 0.374 | 0.373 |
136
+ | 10 | 0.686 | 0.692 | Seed 2.0 Mini | ByteDance | $0.17 | 131 | 256k | 0.441 | 0.250 |
137
137
 
138
138
  ## @task — Subagents: agentic + tool calling, moderate cost sensitivity
139
- eligible: 143 — λ 0.00747 $/quality-point (thinking ×1.857)
139
+ eligible: 146 — λ 0.00747 $/quality-point (thinking ×1.857)
140
140
 
141
141
  | # | value | q | model | org | $/M | tok/s | ctx | ag 29.9% | tool 20.7% | code 18.4% | gen 25.3% | tput 5.7% |
142
142
  | -: | ----: | ----: | --------------------- | ------------------------- | ----: | ----: | ---: | -------: | ---------: | ---------: | --------: | --------: |
143
- | 1 | 0.759 | 0.763 | ★ DeepSeek-V4.1-Flash | DeepSeek | $0.45 | 87 | 1.0M | 0.223 | 0.133 | 0.145 | 0.225 | 0.037 |
144
- | 2 | 0.735 | 0.747 | GLM-5.3 | Zhipu AI | $1.50 | 59 | 1.0M | 0.218 | 0.130 | 0.141 | 0.228 | 0.030 |
145
- | 3 | 0.734 | 0.762 | Muse Spark 1.3 | Meta | $3.71 | 81 | 1.0M | 0.222 | 0.131 | 0.141 | 0.233 | 0.035 |
146
- | 4 | 0.714 | 0.721 | DeepSeek-V4-Pro-0813 | DeepSeek | $0.94 | 49 | 1.0M | 0.210 | 0.123 | 0.136 | 0.224 | 0.027 |
147
- | 5 | 0.713 | 0.720 | MiMo-V2.6-Pro | Xiaomi | $0.99 | 26 | 1.0M | 0.214 | 0.128 | 0.140 | 0.221 | 0.016 |
148
- | 6 | 0.698 | 0.701 | ★ GLM-5.3-Flash | Zhipu AI | $0.36 | 41 | 1.0M | 0.210 | 0.126 | 0.122 | 0.219 | 0.024 |
149
- | 7 | 0.694 | 0.698 | Qwen3.8 Flash | Alibaba Cloud / Qwen Team | $0.43 | 46 | 1.0M | 0.202 | 0.126 | 0.126 | 0.217 | 0.026 |
150
- | 8 | 0.691 | 0.719 | Muse Spark 1.1 | Meta | $3.71 | 170 | 1.0M | 0.201 | 0.127 | 0.124 | 0.219 | 0.048 |
151
- | 9 | 0.683 | 0.704 | Gemini 3.8 Flash | Google | $2.79 | 99 | 1.0M | 0.203 | 0.111 | 0.132 | 0.220 | 0.039 |
152
- | 10 | 0.680 | 0.765 | ★ GPT-5.6 Sol | OpenAI | $11.4 | 60 | 1.1M | 0.225 | 0.126 | 0.149 | 0.234 | 0.030 |
143
+ | 1 | 0.749 | 0.752 | ★ DeepSeek-V4.1-Flash | DeepSeek | $0.42 | 57 | 1.0M | 0.222 | 0.134 | 0.143 | 0.224 | 0.029 |
144
+ | 2 | 0.725 | 0.750 | GLM-5.3 | Zhipu AI | $3.37 | 89 | 1.0M | 0.216 | 0.130 | 0.140 | 0.227 | 0.037 |
145
+ | 3 | 0.724 | 0.752 | Muse Spark 1.3 | Meta | $3.71 | 63 | 1.0M | 0.220 | 0.129 | 0.141 | 0.231 | 0.031 |
146
+ | 4 | 0.713 | 0.720 | MiMo-V2.6-Pro | Xiaomi | $1.01 | 30 | 1.0M | 0.214 | 0.129 | 0.139 | 0.221 | 0.018 |
147
+ | 5 | 0.705 | 0.726 | DeepSeek-V4-Pro-0813 | DeepSeek | $2.79 | 85 | 1.0M | 0.208 | 0.123 | 0.135 | 0.223 | 0.036 |
148
+ | 6 | 0.693 | 0.696 | Qwen3.8 Flash | Alibaba Cloud / Qwen Team | $0.43 | 57 | 1.0M | 0.200 | 0.126 | 0.125 | 0.216 | 0.029 |
149
+ | 7 | 0.692 | 0.695 | ★ GLM-5.3-Flash | Zhipu AI | $0.41 | 37 | 1.0M | 0.207 | 0.125 | 0.122 | 0.218 | 0.022 |
150
+ | 8 | 0.690 | 0.746 | Kimi K3 | Moonshot AI | $7.52 | 58 | 1.0M | 0.216 | 0.131 | 0.141 | 0.228 | 0.030 |
151
+ | 9 | 0.686 | 0.703 | Hy4 preview | Tencent | $2.32 | 33 | — | 0.206 | 0.123 | 0.133 | 0.221 | 0.020 |
152
+ | 10 | 0.684 | 0.712 | Muse Spark 1.1 | Meta | $3.71 | 154 | 1.0M | 0.199 | 0.127 | 0.123 | 0.217 | 0.046 |
153
153
 
154
154
  ## @advisor — Advisor/watchdog: deep reasoning over long context
155
- eligible: 143 — λ 0.01250 $/quality-point (thinking ×1.857)
156
-
157
- | # | value | q | model | org | $/M | tok/s | ctx | rea 43.7% | gen 30.6% | lc 12.6% | tput 13.1% |
158
- | -: | ----: | ----: | -------------------- | ------------------------- | ----: | ----: | ---: | --------: | --------: | -------: | ---------: |
159
- | 1 | 0.788 | 0.834 | ★ Muse Spark 1.3 | Meta | $3.71 | 81 | 1.0M | 0.390 | 0.282 | 0.081 | 0.081 |
160
- | 2 | 0.755 | 0.760 | ★ Qwen3.8 Flash | Alibaba Cloud / Qwen Team | $0.43 | 46 | 1.0M | 0.375 | 0.263 | 0.064 | 0.059 |
161
- | 3 | 0.754 | 0.789 | ★ Gemini 3.8 Flash | Google | $2.79 | 99 | 1.0M | 0.359 | 0.266 | 0.075 | 0.088 |
162
- | 4 | 0.751 | 0.786 | Gemini 3.7 Flash | Google | $2.79 | 99 | 1.0M | 0.358 | 0.259 | 0.080 | 0.088 |
163
- | 5 | 0.750 | 0.756 | DeepSeek-V4.1-Flash | DeepSeek | $0.45 | 87 | 1.0M | 0.375 | 0.272 | 0.024 | 0.083 |
164
- | 6 | 0.739 | 0.758 | GLM-5.3 | Zhipu AI | $1.50 | 59 | 1.0M | 0.390 | 0.276 | 0.024 | 0.068 |
165
- | 7 | 0.733 | 0.779 | Muse Spark 1.1 | Meta | $3.71 | 170 | 1.0M | 0.381 | 0.265 | 0.024 | 0.109 |
166
- | 8 | 0.723 | 0.735 | DeepSeek-V4-Pro-0813 | DeepSeek | $0.94 | 49 | 1.0M | 0.377 | 0.272 | 0.024 | 0.061 |
167
- | 9 | 0.718 | 0.724 | Hy3 | Tencent | $0.47 | 61 | 262k | 0.343 | 0.241 | 0.071 | 0.069 |
168
- | 10 | 0.715 | 0.726 | GPT-5.6 Luna | OpenAI | $0.88 | 57 | 1.1M | 0.347 | 0.248 | 0.065 | 0.067 |
155
+ eligible: 146 — λ 0.01250 $/quality-point (thinking ×1.857)
156
+
157
+ | # | value | q | model | org | $/M | tok/s | ctx | gen 74.3% | lc 12.6% | tput 13.1% |
158
+ | -: | ----: | ----: | --------------------- | ------------------------- | ----: | ----: | ---: | --------: | -------: | ---------: |
159
+ | 1 | 0.786 | 0.833 | ★ Muse Spark 1.3 | Meta | $3.71 | 63 | 1.0M | 0.680 | 0.081 | 0.071 |
160
+ | 2 | 0.761 | 0.766 | ★ Qwen3.8 Flash | Alibaba Cloud / Qwen Team | $0.43 | 57 | 1.0M | 0.635 | 0.064 | 0.067 |
161
+ | 3 | 0.758 | 0.793 | ★ Gemini 3.8 Flash | Google | $2.79 | 78 | 1.0M | 0.639 | 0.075 | 0.079 |
162
+ | 4 | 0.755 | 0.790 | Gemini 3.7 Flash | Google | $2.79 | 91 | 1.0M | 0.624 | 0.080 | 0.085 |
163
+ | 5 | 0.745 | 0.750 | ★ DeepSeek-V4.1-Flash | DeepSeek | $0.42 | 57 | 1.0M | 0.658 | 0.024 | 0.067 |
164
+ | 6 | 0.733 | 0.744 | GPT-5.6 Luna | OpenAI | $0.88 | 83 | 1.1M | 0.598 | 0.065 | 0.082 |
165
+ | 7 | 0.733 | 0.775 | GLM-5.3 | Zhipu AI | $3.37 | 89 | 1.0M | 0.666 | 0.024 | 0.084 |
166
+ | 8 | 0.728 | 0.763 | DeepSeek-V4-Pro-0813 | DeepSeek | $2.79 | 85 | 1.0M | 0.656 | 0.024 | 0.083 |
167
+ | 9 | 0.727 | 0.756 | Hy4 preview | Tencent | $2.32 | 33 | — | 0.649 | 0.061 | 0.045 |
168
+ | 10 | 0.722 | 0.769 | Muse Spark 1.1 | Meta | $3.71 | 154 | 1.0M | 0.639 | 0.024 | 0.105 |
169
169
 
170
170
  ## @designer — Design work: visual/UX judgement on image-capable models
171
- eligible: 88 — λ 0.00882 $/quality-point (thinking ×1.857)
172
-
173
- | # | value | q | model | org | $/M | tok/s | ctx | gen 30.6% | code 11.8% | vis 21.2% | tput 15.3% | web 21.2% | agon |
174
- | -: | ----: | ----: | --------------------- | ----------- | ----: | ----: | ---: | --------: | ---------: | --------: | ---------: | --------: | ---: |
175
- | 1 | 0.787 | 0.791 | ★ DeepSeek-V4.1-Flash | DeepSeek | $0.45 | 87 | 1.0M | 0.272 | 0.092 | 0.134 | 0.097 | 0.195 | — |
176
- | 2 | 0.767 | 0.800 | ★ Muse Spark 1.3 | Meta | $3.71 | 81 | 1.0M | 0.282 | 0.090 | 0.123 | 0.094 | 0.211 | 1289 |
177
- | 3 | 0.761 | 0.785 | Gemini 3.8 Flash | Google | $2.79 | 99 | 1.0M | 0.266 | 0.084 | 0.142 | 0.103 | 0.190 | 1230 |
178
- | 4 | 0.747 | 0.772 | Gemini 3.7 Flash | Google | $2.79 | 99 | 1.0M | 0.259 | 0.079 | 0.138 | 0.103 | 0.193 | 1214 |
179
- | 5 | 0.745 | 0.778 | Muse Spark 1.1 | Meta | $3.71 | 170 | 1.0M | 0.264 | 0.079 | 0.145 | 0.127 | 0.162 | — |
180
- | 6 | 0.720 | 0.729 | MiMo-V2.6-Pro | Xiaomi | $0.99 | 26 | 1.0M | 0.267 | 0.090 | 0.125 | 0.044 | 0.204 | — |
181
- | 7 | 0.708 | 0.711 | ★ GLM-5.3-Flash | Zhipu AI | $0.36 | 41 | 1.0M | 0.265 | 0.078 | 0.135 | 0.063 | 0.169 | — |
182
- | 8 | 0.706 | 0.795 | Kimi K3 | Moonshot AI | $10.1 | 47 | 1.0M | 0.277 | 0.091 | 0.148 | 0.069 | 0.209 | 1285 |
183
- | 9 | 0.703 | 0.803 | ★ GPT-5.6 Sol | OpenAI | $11.4 | 60 | 1.1M | 0.283 | 0.095 | 0.145 | 0.081 | 0.199 | 1183 |
184
- | 10 | 0.698 | 0.722 | Gemini 3.6 Flash | Google | $2.79 | 69 | 1.0M | 0.240 | 0.070 | 0.137 | 0.087 | 0.188 | 1187 |
171
+ eligible: 89 — λ 0.01024 $/quality-point (thinking ×1.857)
172
+
173
+ | # | value | q | model | org | $/M | tok/s | ctx | gen 31.3% | code 7.2% | vis 21.7% | tput 18.1% | web 21.7% | agon |
174
+ | -: | ----: | ----: | --------------------- | ----------- | ----: | ----: | ---: | --------: | --------: | --------: | ---------: | --------: | ---: |
175
+ | 1 | 0.761 | 0.765 | ★ DeepSeek-V4.1-Flash | DeepSeek | $0.42 | 57 | 1.0M | 0.277 | 0.056 | 0.139 | 0.093 | 0.200 | — |
176
+ | 2 | 0.745 | 0.783 | ★ Muse Spark 1.3 | Meta | $3.71 | 63 | 1.0M | 0.287 | 0.055 | 0.127 | 0.098 | 0.216 | 1289 |
177
+ | 3 | 0.741 | 0.770 | ★ Gemini 3.8 Flash | Google | $2.79 | 78 | 1.0M | 0.269 | 0.051 | 0.146 | 0.109 | 0.195 | 1231 |
178
+ | 4 | 0.739 | 0.767 | Gemini 3.7 Flash | Google | $2.79 | 91 | 1.0M | 0.263 | 0.048 | 0.142 | 0.117 | 0.197 | 1214 |
179
+ | 5 | 0.736 | 0.774 | Muse Spark 1.1 | Meta | $3.71 | 154 | 1.0M | 0.269 | 0.048 | 0.148 | 0.145 | 0.163 | — |
180
+ | 6 | 0.721 | 0.798 | ★ Kimi K3 | Moonshot AI | $7.52 | 58 | 1.0M | 0.283 | 0.056 | 0.152 | 0.094 | 0.214 | 1284 |
181
+ | 7 | 0.719 | 0.747 | Gemini 3.6 Flash | Google | $2.79 | 115 | 1.0M | 0.245 | 0.043 | 0.140 | 0.130 | 0.190 | 1187 |
182
+ | 8 | 0.705 | 0.715 | MiMo-V2.6-Pro | Xiaomi | $1.01 | 30 | 1.0M | 0.274 | 0.055 | 0.128 | 0.057 | 0.202 | — |
183
+ | 9 | 0.690 | 0.694 | ★ GLM-5.3-Flash | Zhipu AI | $0.41 | 37 | 1.0M | 0.270 | 0.048 | 0.138 | 0.070 | 0.168 | — |
184
+ | 10 | 0.683 | 0.764 | GPT-6.1 Sol | OpenAI | $7.84 | 39 | 1.1M | 0.291 | 0.056 | 0.152 | 0.072 | 0.193 | — |
185
+ focus: website — coverage 123/407 (30.2%) ok; dispersion 0.992 ok; composition ok; freshness 0.3mo ok; trust unknown; modality unknown; maintenance unknown; provenance unknown; cross-source unknown
185
186
 
186
187
  # Suggested settings.modelRoles (best-fit #1 per role, resolved via the omp catalog).
187
188
  # Selectors are openrouter/<id>; the omp-llm-role plugin writes this block daily.
188
189
  modelRoles:
189
190
  default: "openrouter/deepseek/deepseek-v4.1-flash"
190
191
  smol: "openrouter/meta/muse-spark-1.1"
191
- slow: "openrouter/z-ai/glm-5.3"
192
+ slow: "openrouter/meta/muse-spark-1.3"
192
193
  vision: "openrouter/qwen/qwen3.8-flash"
193
- plan: "openrouter/qwen/qwen3.8-flash"
194
+ plan: "openrouter/meta/muse-spark-1.3"
194
195
  commit: "openrouter/meta/muse-spark-1.1"
195
- tiny: "openrouter/meta/muse-spark-1.1"
196
+ tiny: "openrouter/inception/mercury-2"
196
197
  task: "openrouter/deepseek/deepseek-v4.1-flash"
197
198
  advisor: "openrouter/meta/muse-spark-1.3"
198
199
  designer: "openrouter/deepseek/deepseek-v4.1-flash"