@oonacode/linux-x64 0.3.302 → 0.3.316

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.
package/oona CHANGED
@@ -1,4 +1,4 @@
1
1
  [diffend] Oversized file quarantined before diffing.
2
2
  name: package/oona
3
- size: 108656096 bytes
4
- sha256: 748c122a1c6c0869582ddf3dad782b397336f04b07982a2245a9866e6b8c7c63
3
+ size: 108897760 bytes
4
+ sha256: 50b90a64900113b8e8078bf269b01fcbeec53da34ea3d934d1a2ce4d3bd9d42f
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@oonacode/linux-x64",
3
- "version": "0.3.302",
3
+ "version": "0.3.316",
4
4
  "description": "OonaCode native binaries for Linux x64: oona (CLI + headless host service), the pinned oonaclaude engine sidecar, local web UI assets. Installed automatically as an optional dependency of `oonacode`.",
5
5
  "license": "UNLICENSED",
6
6
  "homepage": "https://oonacode.oonak.ai",
package/registry.json CHANGED
@@ -600,7 +600,7 @@
600
600
  "poolOnly": true,
601
601
  "contributorMode": {
602
602
  "suffix": "-contributor",
603
- "summary": "Meta lists every Muse model twice; the -contributor twin is cheaper, and in exchange Meta may use the content sent to it for product improvement and to train its models."
603
+ "summary": "Some Muse models have a cheaper -contributor twin; in exchange Meta may use the content sent to that twin for product improvement and to train its models. Muse Ember has no Contributor-mode variant."
604
604
  },
605
605
  "baseUrl": "https://api.meta.ai/v1",
606
606
  "messagesPath": "/responses",
@@ -613,6 +613,18 @@
613
613
  },
614
614
  "defaultModel": "muse-spark-1.3",
615
615
  "models": [
616
+ {
617
+ "id": "muse-ember-1.0",
618
+ "displayName": "Muse Ember 1.0",
619
+ "contextWindow": 1048576,
620
+ "supportsTools": true,
621
+ "supportsThinking": true,
622
+ "supportsVision": true,
623
+ "effortLevels": ["low", "medium", "high", "xhigh", "max"],
624
+ "defaultEffort": "high",
625
+ "maxOutputTokens": 524288,
626
+ "notes": "Meta's Muse Ember guide lists 1,048,576 context, 524,288 output, tool calling, structured output and image/PDF input; available in Muse Code pay-as-you-go and every plan, with no Contributor-mode variant (dev.meta.ai/docs/muse-ember, read 2026-11-11). Standard id only; NOT pooled, and the admin refuses to pool it: the pool serves only contributor twins (owner, 2026-11-04: Meta may use what is sent to a contributor model, and he accepted that for those four ids only) and Ember has none. It runs on a person's own Muse plan. API efforts minimal through max (none is refused); OonaCode deliberately keeps high like its other Muse models, while Meta documents max as the API default. Measured offline against the real Muse Code 1.4.15 (canned backend, 2026-11-11): with contributor mode on, Ember is sent unchanged with header 0, and started without an effort flag the CLI asks for max. Meta's list price per 1M tokens at <=200K input: $0.10 input / $0.004 cached / $0.50 output; above 200K (cached input included): $0.30 / $0.012 / $1.50 for the whole request. On a subscription, Meta says requests above 200K use the plan's allowance faster, without specifying a multiplier. No live account catalogue or live Ember turn was checked (the 1.4.15 check was offline, against a canned backend)."
627
+ },
616
628
  {
617
629
  "id": "muse-spark-1.3",
618
630
  "displayName": "Muse Spark 1.3",
@@ -781,6 +793,18 @@
781
793
  }
782
794
  },
783
795
  "models": [
796
+ {
797
+ "id": "muse-ember-1.0",
798
+ "displayName": "Muse Ember 1.0",
799
+ "contextWindow": 1048576,
800
+ "supportsTools": true,
801
+ "supportsThinking": true,
802
+ "supportsVision": true,
803
+ "effortLevels": ["low", "medium", "high", "xhigh", "max"],
804
+ "defaultEffort": "high",
805
+ "maxOutputTokens": 524288,
806
+ "notes": "The person's own Muse key, used directly from their PC. Meta documents Muse Ember for Muse Code pay-as-you-go and every plan, with no Contributor-mode variant: 1,048,576 context, 524,288 output, tool calling, structured output and image/PDF input (dev.meta.ai/docs/muse-ember, read 2026-11-11). API efforts minimal through max (none is refused); OonaCode deliberately keeps high like its other Muse models, while Meta documents max as the API default. Measured offline against the real Muse Code 1.4.15 (canned backend, 2026-11-11): with contributor mode on, Ember is sent unchanged with header 0, and started without an effort flag the CLI asks for max. No live account catalogue or live Ember turn was checked. On a subscription, requests above 200K input use the plan's allowance faster; Meta gives no multiplier."
807
+ },
784
808
  {
785
809
  "id": "muse-spark-1.3",
786
810
  "displayName": "Muse Spark 1.3",
@@ -900,6 +924,370 @@
900
924
  "notes": "The user's OWN Meta Muse key — a Muse Code subscription key (minted by `muse login`) or a Meta Model API key — kept in the vault and used straight from the PC (guardrail 4; the gateway is never on the path and this entry seeds no server rows). Muse speaks the OpenAI Responses API at api.meta.ai/v1/responses, so the engine reaches it through the host's loopback upstream, which translates every request (translate.wire = responses) and presents Muse's identity headers. Meta's terms: a subscription key is for use with Muse Code only — that is the person's standing, as for the other consumer plans. Models and limits are what Meta's own model list answered on 2026-10-13. docs/muse-engine.md.",
901
925
  "defaultEffort": "high"
902
926
  },
927
+ {
928
+ "id": "github-copilot",
929
+ "displayName": "GitHub Copilot (Copilot plan)",
930
+ "effortLevels": ["low", "medium", "high", "xhigh", "max"],
931
+ "kind": "openai-compatible",
932
+ "route": "gateway",
933
+ "credentialClass": "api_key",
934
+ "poolOnly": true,
935
+ "baseUrl": "https://api.individual.githubcopilot.com",
936
+ "messagesPath": "/responses",
937
+ "auth": {
938
+ "header": "authorization-bearer",
939
+ "extraHeaders": {
940
+ "copilot-integration-id": "copilot-developer-cli",
941
+ "copilot-harness-id": "copilot-sdk",
942
+ "editor-version": "copilot/1.0.107",
943
+ "x-github-api-version": "2026-08-01",
944
+ "user-agent": "copilot/1.0.107 (win32 v24.22.1) term/unknown client/github/cli"
945
+ }
946
+ },
947
+ "defaultModel": "copilot-auto",
948
+ "models": [
949
+ {
950
+ "id": "copilot-auto",
951
+ "displayName": "Copilot Auto",
952
+ "contextWindow": 128000,
953
+ "supportsTools": true,
954
+ "supportsThinking": true,
955
+ "supportsVision": true,
956
+ "effortLevels": ["low", "medium", "high"],
957
+ "maxOutputTokens": 128000,
958
+ "notes": "GitHub picks the model your plan allows (POST /auto) and the chat keeps it for 24 hours. The only way to GPT-6 Luna, GPT-5.6 Luna and MAI-Code-1.1-Flash on a Free plan, where naming them is refused. Metered at the dearest of those (GitHub's catalogue price $0.20 input / $0.02 cached / $1.20 output per 1M tokens (billing.token_prices of GET /models, 2026-11-07; Copilot bills AI credits, 100 credits = $1).) because the real model is GitHub's choice. Window 128k = the smallest model Auto may choose.",
959
+ "defaultEffort": "medium"
960
+ },
961
+ {
962
+ "id": "gpt-5.4-nano",
963
+ "displayName": "GPT-5.4 nano",
964
+ "contextWindow": 272000,
965
+ "supportsTools": true,
966
+ "supportsThinking": true,
967
+ "supportsVision": true,
968
+ "effortLevels": ["low", "medium", "high", "xhigh"],
969
+ "maxOutputTokens": 128000,
970
+ "notes": "Included in every plan, Free too, and costs no Copilot credits (price 0 in the catalogue). Answered by name on a Free plan (tested 2026-11-07).",
971
+ "defaultEffort": "low"
972
+ },
973
+ {
974
+ "id": "gpt-6-luna",
975
+ "displayName": "GPT-6 Luna",
976
+ "contextWindow": 872000,
977
+ "supportsTools": true,
978
+ "supportsThinking": true,
979
+ "supportsVision": true,
980
+ "effortLevels": ["low", "medium", "high", "xhigh", "max"],
981
+ "maxOutputTokens": 128000,
982
+ "notes": "Enabled on Free. By name only on a paid plan; on Free reach it through Copilot Auto. GitHub's catalogue price $0.10 input / $0.01 cached / $0.50 output per 1M tokens (billing.token_prices of GET /models, 2026-11-07; Copilot bills AI credits, 100 credits = $1).",
983
+ "defaultEffort": "medium"
984
+ },
985
+ {
986
+ "id": "gpt-5.6-luna",
987
+ "displayName": "GPT-5.6 Luna",
988
+ "contextWindow": 922000,
989
+ "supportsTools": true,
990
+ "supportsThinking": true,
991
+ "supportsVision": true,
992
+ "effortLevels": ["low", "medium", "high", "xhigh", "max"],
993
+ "maxOutputTokens": 128000,
994
+ "notes": "Enabled on Free. By name only on a paid plan; on Free reach it through Copilot Auto. GitHub's catalogue price $0.20 input / $0.02 cached / $1.20 output per 1M tokens (billing.token_prices of GET /models, 2026-11-07; Copilot bills AI credits, 100 credits = $1).",
995
+ "defaultEffort": "medium"
996
+ },
997
+ {
998
+ "id": "gpt-5.6-terra",
999
+ "displayName": "GPT-5.6 Terra",
1000
+ "contextWindow": 922000,
1001
+ "supportsTools": true,
1002
+ "supportsThinking": true,
1003
+ "supportsVision": true,
1004
+ "effortLevels": ["low", "medium", "high", "xhigh", "max"],
1005
+ "maxOutputTokens": 128000,
1006
+ "notes": "Copilot Pro and above. NOT verified: both of the owner's accounts are Free. GitHub's catalogue price $2 input / $0.20 cached / $12 output per 1M tokens (billing.token_prices of GET /models, 2026-11-07; Copilot bills AI credits, 100 credits = $1).",
1007
+ "defaultEffort": "medium"
1008
+ },
1009
+ {
1010
+ "id": "gpt-5.6-sol",
1011
+ "displayName": "GPT-5.6 Sol",
1012
+ "contextWindow": 922000,
1013
+ "supportsTools": true,
1014
+ "supportsThinking": true,
1015
+ "supportsVision": true,
1016
+ "effortLevels": ["low", "medium", "high", "xhigh", "max"],
1017
+ "maxOutputTokens": 128000,
1018
+ "notes": "Copilot Pro+ and above. NOT verified (Free accounts only). GitHub's catalogue price $4 input / $0.40 cached / $20 output per 1M tokens (billing.token_prices of GET /models, 2026-11-07; Copilot bills AI credits, 100 credits = $1).",
1019
+ "defaultEffort": "medium"
1020
+ },
1021
+ {
1022
+ "id": "gpt-6-sol",
1023
+ "displayName": "GPT-6 Sol",
1024
+ "contextWindow": 872000,
1025
+ "supportsTools": true,
1026
+ "supportsThinking": true,
1027
+ "supportsVision": true,
1028
+ "effortLevels": ["low", "medium", "high", "xhigh", "max"],
1029
+ "maxOutputTokens": 128000,
1030
+ "notes": "Copilot Pro+ and above. NOT verified (Free accounts only). GitHub's catalogue price $2 input / $0.20 cached / $10 output per 1M tokens (billing.token_prices of GET /models, 2026-11-07; Copilot bills AI credits, 100 credits = $1).",
1031
+ "defaultEffort": "medium"
1032
+ },
1033
+ {
1034
+ "id": "gpt-6.1-sol",
1035
+ "displayName": "GPT-6.1 Sol",
1036
+ "contextWindow": 922000,
1037
+ "supportsTools": true,
1038
+ "supportsThinking": true,
1039
+ "supportsVision": true,
1040
+ "effortLevels": ["low", "medium", "high", "xhigh", "max"],
1041
+ "maxOutputTokens": 128000,
1042
+ "notes": "Copilot Pro+ and above. NOT verified (Free accounts only). GitHub's catalogue price $2 input / $0.10 cached / $10 output per 1M tokens (billing.token_prices of GET /models, 2026-11-07; Copilot bills AI credits, 100 credits = $1).",
1043
+ "defaultEffort": "medium"
1044
+ },
1045
+ {
1046
+ "id": "gpt-6-astra",
1047
+ "displayName": "GPT-6 Astra",
1048
+ "contextWindow": 1050000,
1049
+ "supportsTools": true,
1050
+ "supportsThinking": true,
1051
+ "supportsVision": true,
1052
+ "effortLevels": ["low", "medium", "high", "xhigh", "max"],
1053
+ "maxOutputTokens": 128000,
1054
+ "notes": "Copilot Pro+ and above. NOT verified (Free accounts only). GitHub's catalogue price $10 input / $1 cached / $50 output per 1M tokens (billing.token_prices of GET /models, 2026-11-07; Copilot bills AI credits, 100 credits = $1).",
1055
+ "defaultEffort": "medium"
1056
+ }
1057
+ ],
1058
+ "modelAliases": {
1059
+ "opus": "copilot-auto",
1060
+ "sonnet": "copilot-auto",
1061
+ "haiku": "copilot-auto"
1062
+ },
1063
+ "rateLimits": {
1064
+ "maxConcurrent": 2,
1065
+ "requestsPerMinute": 30
1066
+ },
1067
+ "capabilities": {
1068
+ "streaming": true,
1069
+ "tools": true,
1070
+ "thinking": true,
1071
+ "vision": true,
1072
+ "countTokens": false,
1073
+ "promptCaching": true
1074
+ },
1075
+ "translate": {
1076
+ "wire": "responses",
1077
+ "systemRole": "developer",
1078
+ "reasoningEffortFromBudget": true,
1079
+ "supportsParallelToolCalls": true
1080
+ },
1081
+ "phase": 1,
1082
+ "enabled": true,
1083
+ "recordLabel": "Subscription",
1084
+ "docsUrl": "https://docs.github.com/en/copilot",
1085
+ "references": [
1086
+ {
1087
+ "label": "GitHub Copilot CLI",
1088
+ "url": "https://github.com/github/copilot-cli",
1089
+ "kind": "docs"
1090
+ },
1091
+ {
1092
+ "label": "GitHub Copilot plans and models",
1093
+ "url": "https://docs.github.com/en/copilot/get-started/plans",
1094
+ "kind": "pricing"
1095
+ },
1096
+ {
1097
+ "label": "GitHub Copilot terms",
1098
+ "url": "https://docs.github.com/en/site-policy/github-terms/github-terms-for-additional-products-and-features#github-copilot",
1099
+ "kind": "other"
1100
+ }
1101
+ ],
1102
+ "notes": "OPERATOR POOL ONLY (owner, 2026-11-07: \"integrate that last loggedin account on server and show usage in dashboard\"). The operator's GitHub Copilot plan, signed in from Admin -> Providers -> GitHub Copilot with GitHub's device flow (the Copilot CLI's public client id, scope read:user ONLY). The token is a plain bearer that never expires and has no refresh token. The engine's request is translated to the Responses wire and sent to the account's api host with the Copilot CLI's identity headers (translate/copilot-dialect.ts). Use a DEDICATED GitHub account for the pool, never the owner's own: GitHub can suspend an account that serves other people's traffic. Claude, Gemini and Kimi models are served by Copilot on other wires (/v1/messages, /chat/completions) and are not built yet; paid models are untested (both accounts the owner tried are Free). Design: docs/github-copilot.md.",
1103
+ "credentialVariants": [
1104
+ {
1105
+ "id": "token-plan",
1106
+ "label": "GitHub Copilot plan (sign in)",
1107
+ "header": "authorization-bearer",
1108
+ "operatorOnly": true,
1109
+ "baseUrl": "https://api.individual.githubcopilot.com",
1110
+ "subscriptionLogin": {
1111
+ "type": "oauth-device",
1112
+ "deviceAuthorizationUrl": "https://github.com/login/device/code",
1113
+ "tokenUrl": "https://github.com/login/oauth/access_token",
1114
+ "clientId": "Ov23ctDVkRmgkPke0Mmm",
1115
+ "scopes": ["read:user"],
1116
+ "signInLabel": "Sign in with GitHub (Copilot)",
1117
+ "tokensNeverExpire": true,
1118
+ "identify": "github-copilot"
1119
+ },
1120
+ "notes": "One GitHub account's Copilot plan. The sign-in reads GET api.github.com/copilot_internal/user to name the plan and the api host, and refuses an account with no Copilot plan. Usage (the plan's AI credits) is read from the same endpoint."
1121
+ }
1122
+ ],
1123
+ "defaultEffort": "medium"
1124
+ },
1125
+ {
1126
+ "id": "github-copilot-subscription",
1127
+ "displayName": "GitHub Copilot Subscription",
1128
+ "effortLevels": ["low", "medium", "high", "xhigh", "max"],
1129
+ "kind": "openai-compatible",
1130
+ "route": "direct",
1131
+ "credentialClass": "subscription",
1132
+ "baseUrl": "https://api.individual.githubcopilot.com",
1133
+ "messagesPath": "/responses",
1134
+ "defaultModel": "copilot-auto",
1135
+ "auth": {
1136
+ "header": "authorization-bearer",
1137
+ "extraHeaders": {
1138
+ "copilot-integration-id": "copilot-developer-cli",
1139
+ "copilot-harness-id": "copilot-sdk",
1140
+ "editor-version": "copilot/1.0.107",
1141
+ "x-github-api-version": "2026-08-01",
1142
+ "user-agent": "copilot/1.0.107 (win32 v24.22.1) term/unknown client/github/cli"
1143
+ }
1144
+ },
1145
+ "models": [
1146
+ {
1147
+ "id": "copilot-auto",
1148
+ "displayName": "Copilot Auto",
1149
+ "contextWindow": 128000,
1150
+ "supportsTools": true,
1151
+ "supportsThinking": true,
1152
+ "supportsVision": true,
1153
+ "effortLevels": ["low", "medium", "high"],
1154
+ "maxOutputTokens": 128000,
1155
+ "notes": "GitHub picks the model your plan allows (POST /auto) and the chat keeps it for 24 hours. The only way to GPT-6 Luna, GPT-5.6 Luna and MAI-Code-1.1-Flash on a Free plan, where naming them is refused. Metered at the dearest of those (GitHub's catalogue price $0.20 input / $0.02 cached / $1.20 output per 1M tokens (billing.token_prices of GET /models, 2026-11-07; Copilot bills AI credits, 100 credits = $1).) because the real model is GitHub's choice. Window 128k = the smallest model Auto may choose.",
1156
+ "defaultEffort": "medium"
1157
+ },
1158
+ {
1159
+ "id": "gpt-5.4-nano",
1160
+ "displayName": "GPT-5.4 nano",
1161
+ "contextWindow": 272000,
1162
+ "supportsTools": true,
1163
+ "supportsThinking": true,
1164
+ "supportsVision": true,
1165
+ "effortLevels": ["low", "medium", "high", "xhigh"],
1166
+ "maxOutputTokens": 128000,
1167
+ "notes": "Included in every plan, Free too, and costs no Copilot credits (price 0 in the catalogue). Answered by name on a Free plan (tested 2026-11-07).",
1168
+ "defaultEffort": "low"
1169
+ },
1170
+ {
1171
+ "id": "gpt-6-luna",
1172
+ "displayName": "GPT-6 Luna",
1173
+ "contextWindow": 872000,
1174
+ "supportsTools": true,
1175
+ "supportsThinking": true,
1176
+ "supportsVision": true,
1177
+ "effortLevels": ["low", "medium", "high", "xhigh", "max"],
1178
+ "maxOutputTokens": 128000,
1179
+ "notes": "Enabled on Free. By name only on a paid plan; on Free reach it through Copilot Auto. GitHub's catalogue price $0.10 input / $0.01 cached / $0.50 output per 1M tokens (billing.token_prices of GET /models, 2026-11-07; Copilot bills AI credits, 100 credits = $1).",
1180
+ "defaultEffort": "medium"
1181
+ },
1182
+ {
1183
+ "id": "gpt-5.6-luna",
1184
+ "displayName": "GPT-5.6 Luna",
1185
+ "contextWindow": 922000,
1186
+ "supportsTools": true,
1187
+ "supportsThinking": true,
1188
+ "supportsVision": true,
1189
+ "effortLevels": ["low", "medium", "high", "xhigh", "max"],
1190
+ "maxOutputTokens": 128000,
1191
+ "notes": "Enabled on Free. By name only on a paid plan; on Free reach it through Copilot Auto. GitHub's catalogue price $0.20 input / $0.02 cached / $1.20 output per 1M tokens (billing.token_prices of GET /models, 2026-11-07; Copilot bills AI credits, 100 credits = $1).",
1192
+ "defaultEffort": "medium"
1193
+ },
1194
+ {
1195
+ "id": "gpt-5.6-terra",
1196
+ "displayName": "GPT-5.6 Terra",
1197
+ "contextWindow": 922000,
1198
+ "supportsTools": true,
1199
+ "supportsThinking": true,
1200
+ "supportsVision": true,
1201
+ "effortLevels": ["low", "medium", "high", "xhigh", "max"],
1202
+ "maxOutputTokens": 128000,
1203
+ "notes": "Copilot Pro and above. NOT verified: both of the owner's accounts are Free. GitHub's catalogue price $2 input / $0.20 cached / $12 output per 1M tokens (billing.token_prices of GET /models, 2026-11-07; Copilot bills AI credits, 100 credits = $1).",
1204
+ "defaultEffort": "medium"
1205
+ },
1206
+ {
1207
+ "id": "gpt-5.6-sol",
1208
+ "displayName": "GPT-5.6 Sol",
1209
+ "contextWindow": 922000,
1210
+ "supportsTools": true,
1211
+ "supportsThinking": true,
1212
+ "supportsVision": true,
1213
+ "effortLevels": ["low", "medium", "high", "xhigh", "max"],
1214
+ "maxOutputTokens": 128000,
1215
+ "notes": "Copilot Pro+ and above. NOT verified (Free accounts only). GitHub's catalogue price $4 input / $0.40 cached / $20 output per 1M tokens (billing.token_prices of GET /models, 2026-11-07; Copilot bills AI credits, 100 credits = $1).",
1216
+ "defaultEffort": "medium"
1217
+ },
1218
+ {
1219
+ "id": "gpt-6-sol",
1220
+ "displayName": "GPT-6 Sol",
1221
+ "contextWindow": 872000,
1222
+ "supportsTools": true,
1223
+ "supportsThinking": true,
1224
+ "supportsVision": true,
1225
+ "effortLevels": ["low", "medium", "high", "xhigh", "max"],
1226
+ "maxOutputTokens": 128000,
1227
+ "notes": "Copilot Pro+ and above. NOT verified (Free accounts only). GitHub's catalogue price $2 input / $0.20 cached / $10 output per 1M tokens (billing.token_prices of GET /models, 2026-11-07; Copilot bills AI credits, 100 credits = $1).",
1228
+ "defaultEffort": "medium"
1229
+ },
1230
+ {
1231
+ "id": "gpt-6.1-sol",
1232
+ "displayName": "GPT-6.1 Sol",
1233
+ "contextWindow": 922000,
1234
+ "supportsTools": true,
1235
+ "supportsThinking": true,
1236
+ "supportsVision": true,
1237
+ "effortLevels": ["low", "medium", "high", "xhigh", "max"],
1238
+ "maxOutputTokens": 128000,
1239
+ "notes": "Copilot Pro+ and above. NOT verified (Free accounts only). GitHub's catalogue price $2 input / $0.10 cached / $10 output per 1M tokens (billing.token_prices of GET /models, 2026-11-07; Copilot bills AI credits, 100 credits = $1).",
1240
+ "defaultEffort": "medium"
1241
+ },
1242
+ {
1243
+ "id": "gpt-6-astra",
1244
+ "displayName": "GPT-6 Astra",
1245
+ "contextWindow": 1050000,
1246
+ "supportsTools": true,
1247
+ "supportsThinking": true,
1248
+ "supportsVision": true,
1249
+ "effortLevels": ["low", "medium", "high", "xhigh", "max"],
1250
+ "maxOutputTokens": 128000,
1251
+ "notes": "Copilot Pro+ and above. NOT verified (Free accounts only). GitHub's catalogue price $10 input / $1 cached / $50 output per 1M tokens (billing.token_prices of GET /models, 2026-11-07; Copilot bills AI credits, 100 credits = $1).",
1252
+ "defaultEffort": "medium"
1253
+ }
1254
+ ],
1255
+ "rateLimits": {
1256
+ "maxConcurrent": 2,
1257
+ "requestsPerMinute": 30
1258
+ },
1259
+ "capabilities": {
1260
+ "streaming": true,
1261
+ "tools": true,
1262
+ "thinking": true,
1263
+ "vision": true,
1264
+ "countTokens": false,
1265
+ "promptCaching": true
1266
+ },
1267
+ "translate": {
1268
+ "wire": "responses",
1269
+ "systemRole": "developer",
1270
+ "reasoningEffortFromBudget": true,
1271
+ "supportsParallelToolCalls": true
1272
+ },
1273
+ "phase": 1,
1274
+ "enabled": true,
1275
+ "adminGroup": "github-copilot",
1276
+ "recordLabel": "Subscription",
1277
+ "subscriptionLogin": {
1278
+ "type": "oauth-device",
1279
+ "deviceAuthorizationUrl": "https://github.com/login/device/code",
1280
+ "tokenUrl": "https://github.com/login/oauth/access_token",
1281
+ "clientId": "Ov23ctDVkRmgkPke0Mmm",
1282
+ "scopes": ["read:user"],
1283
+ "signInLabel": "Sign in with GitHub (Copilot)",
1284
+ "tokensNeverExpire": true,
1285
+ "identify": "github-copilot"
1286
+ },
1287
+ "docsUrl": "https://docs.github.com/en/copilot",
1288
+ "notes": "The user's OWN GitHub Copilot plan (Free, Pro, Pro+, Max, Business), signed in with GitHub's device flow ON THE PC (the Copilot CLI's public client id, scope read:user only - a leaked token cannot reach a repository). The token is stored in the vault and the PC translates the engine's request to Copilot's Responses wire itself (engine/local-upstream.ts): no OonaCode server sees the token or the chat (guardrails 4 and 5). Claude, Gemini and Kimi models need other wires and are not built yet. Design: docs/github-copilot.md.",
1289
+ "defaultEffort": "medium"
1290
+ },
903
1291
  {
904
1292
  "id": "xai",
905
1293
  "displayName": "xAI (Grok)",
@@ -1756,6 +2144,8 @@
1756
2144
  "supportsVision": true,
1757
2145
  "inputPricePerM": 0.75,
1758
2146
  "outputPricePerM": 3.75,
2147
+ "effortLevels": ["low", "medium", "high"],
2148
+ "defaultEffort": "high",
1759
2149
  "notes": "Google lists one model per level — gemini-3.8-flash-high, -medium and -low (fetchAvailableModels, 2026-10-03: 1,048,576 context / 65,536 output each); OonaCode shows ONE model and the chat’s effort picker fills the level (upstreamId template; xhigh and max read as high; no effort = high; agy 1.5.0 accepts --effort low, medium or high only and rejects xhigh and max for every model). Asked as agy 1.5.0 asks it: thoughts included and the id's level as thinkingLevel LOW / MEDIUM / HIGH, no budget (agy up to 1.3.x sent Low 1000, Medium 4000, High -1). Cost = Google’s Gemini API list price for gemini-3.8-flash: $0.75 in / $0.075 cached / $3.75 out per 1M through 2026-12-31, DOUBLING to $1.50 / $0.15 / $7.50 from 2027-01-01 (ai.google.dev/gemini-api/docs/pricing, updated 2026-10-01) — move the cost cells on the model page then."
1760
2150
  },
1761
2151
  {
@@ -1769,8 +2159,25 @@
1769
2159
  "supportsVision": true,
1770
2160
  "inputPricePerM": 2,
1771
2161
  "outputPricePerM": 10,
2162
+ "effortLevels": ["low", "medium", "high"],
2163
+ "defaultEffort": "high",
1772
2164
  "notes": "Google lists one model per level - gemini-4-argon-high, -medium and -low (fetchAvailableModels on the operator's Ultra account, 2026-10-13: 1,048,576 context / 65,536 output each; model enums M417 / M418 / M419); OonaCode shows ONE model and the chat's effort picks the level. Asked like the Google-hosted Claude models: thoughts included and the id's level as thinkingLevel, no budget (agy 1.4.0 and 1.5.0; agy 1.3.3 sent a dynamic budget of -1 beside it). Ultra accounts only: a non-Ultra personal account does not list it. Google's Gemini API list price is INTRODUCTORY $2 / $0.10 cached / $10 per 1M through 2026-11-30 and $4 / $0.20 / $20 from 2026-12-01 - move the catalogue cells then."
1773
2165
  },
2166
+ {
2167
+ "id": "gemini-4-carbon",
2168
+ "displayName": "Gemini 4 Carbon",
2169
+ "upstreamId": "gemini-4-carbon-{effort:high}",
2170
+ "contextWindow": 1048576,
2171
+ "maxOutputTokens": 262144,
2172
+ "supportsTools": true,
2173
+ "supportsThinking": true,
2174
+ "supportsVision": true,
2175
+ "inputPricePerM": 4,
2176
+ "outputPricePerM": 20,
2177
+ "effortLevels": ["low", "medium", "high"],
2178
+ "defaultEffort": "high",
2179
+ "notes": "Google lists one model per level - gemini-4-carbon-high, -medium and -low (fetchAvailableModels, 2026-11-10: 1,048,576 context / 262,144 output each; model enums M427 / M437 / M438); OonaCode shows ONE model and the chat's effort picks the level. Asked as agy 1.5.x asks it: thoughts included and the id's level as thinkingLevel LOW / MEDIUM / HIGH, no budget. Price set by the owner, 2026-11-11: $4 per 1M input, $20 per 1M output, 95 % cache discount (cached input $0.20 per 1M), flat - no 2026-12-01 step."
2180
+ },
1774
2181
  {
1775
2182
  "id": "google-claude-opus-5-5",
1776
2183
  "displayName": "Claude Opus 5.5 (via Google)",
@@ -1851,7 +2258,7 @@
1851
2258
  "operatorOnly": true,
1852
2259
  "baseUrl": "https://daily-cloudcode-pa.googleapis.com",
1853
2260
  "extraHeaders": {
1854
- "user-agent": "antigravity/cli/1.5.0 (aidev_client; os_type=linux; arch=amd64; cl=1013440158; auth_method=consumer)"
2261
+ "user-agent": "antigravity/cli/1.5.5 (aidev_client; os_type=linux; arch=amd64; cl=1017970541; auth_method=consumer)"
1855
2262
  },
1856
2263
  "agyLogin": {
1857
2264
  "type": "agy-server",