[model-inventory] Model alias inventory update - 2026-10-07
维护者通常 1 天内回复
评估
- 难度
- 2/5
- 预计耗时
- 1-3 小时
- 新手友好度
- 72/100
- Issue 类型
- 缺陷
- 描述清晰度
- 描述清楚
- 活跃度
- 活跃
- 技术栈
- github-actions, json
- 领域
- data
调研方向
从 pkg/cli/data/models.json 和 actions/setup/js/models.json 开始；issue 称这两个文件完全相同。将其中的模型记录和价格与报告的 Copilot SDK 值进行比较。按建议运行 make refresh-models-json，然后调整文件以添加 claude-haiku-5.5 并修正列出的 cache_read 差异。两个文件一致且报告的价格正确时，即为完成。
由索引模型根据 Issue 内容生成。
描述
Summary
4 pricing items + 1 stale alias found; no new model families lacking alias coverage.
- Providers queried: OpenAI (135), Anthropic (12), Gemini (50), Copilot SDK (24), Copilot reflect (42 IDs, incl. deprecated/embedding IDs)
- Proposed alias changes: 1 (optional cleanup)
- Pricing gaps found: 5
- Docs pricing table (Step 2.5): skipped — Playwright extraction was not performed; the billing artifact is an empty placeholder. SDK vs reflect used for validation.
Provider Model Counts
| Provider | Models | Status |
|---|---|---|
| openai | 135 | ✅ ok |
| anthropic | 12 | ✅ ok |
| gemini | 50 | ✅ ok |
| copilot-sdk | 24 | ✅ ok |
| copilot (reflect) | 42 | ✅ ok |
Raw API Fields Discovered
- Copilot SDK:
billing.tokenPrices(inputPrice, outputPrice, cacheReadPrice, cacheWritePrice, contextMax, longContext tier),discountPercent(auto only). - Reflect: model ID list only (no billing multiplier in this run).
- Gemini: input/output token limits; Anthropic:
display_name; OpenAI:owned_by.
models.json Pricing Analysis
(pkg/cli/data/models.json and actions/setup/js/models.json are identical.)
Missing from models.json
| Model ID | Provider | Inferred Pricing (per 1M tokens) | Basis |
|---|---|---|---|
| claude-haiku-5.5 | copilot / anthropic (claude-haiku-5-5) |
in $10 (SDK units: 10, out 50, cache read 1, cache write 12.5) → stored form: input 1e-07, output 5e-07, cache_read 1e-08, cache_write 1.25e-07 | Copilot SDK billing.tokenPrices |
Note: models.json stores per-token costs; SDK prices are in units of 1e-8 $/token (e.g. 200 → 2e-06). provider_type would be anthropic.
Historical entries not currently returned
Kept as historical: claude-fable-5, claude-fable-5.1, claude-opus-4.5/4.6/4.7, claude-sonnet-4/4.5/4.6, gemini-2.5-pro, gemini-3-flash-preview, gemini-3.1-pro-preview, gemini-3.5/3.6/3.7/3.8-flash (not in Copilot SDK; present in Gemini API), gpt-5.2, gpt-5.2-codex, kimi-k2.7-code, kimi-k3, mai-code-1-flash-picker, raptor-mini.
Inferred vs stored pricing discrepancies
| Model ID | Stored | SDK | Field |
|---|---|---|---|
| claude-sonnet-5.5 | cache_read 2e-07 | 1e-07 | cache_read |
| gpt-5.4-mini | cache_read 7e-08 | 7.5e-08 | cache_read (rounding) |
| gpt-5.3-codex | cache_read 1.7e-07 | 1.75e-07 | cache_read (rounding) |
| gpt-5-mini | cache_read 2e-08 | 2.5e-08 | cache_read (rounding) |
All other input/output/cache prices of the 23 SDK-priced models match.
Proposed Alias Updates
- What: Optional cleanup of
reasoningalias. Why: no live Copilot model matchescopilot/o1*|o3*|o4*; OpenAI API still lists o1/o3/o4-mini, so no change is required — keep as is. - All live models are covered by existing aliases (
haikucovers claude-haiku-5.5;sonnet-6xcovers sonnet-5.5;gpt-6covers gpt-6.1-sol/astra/luna/sol;gemini-flash-litecovers gemini-3.8-flash-lite-*;grokcovers 4.5–4.7;mai-codecovers mai-code-1.1-flash). - Task-oriented aliases (
coding,reasoning,vision,summarization) already exist.
Full Copilot Model IDs (reflect)
claude-haiku-4.5, claude-haiku-5.5, claude-opus-4.8, claude-opus-5, claude-opus-5.5, claude-sonnet-5, claude-sonnet-5.5, gpt-5-mini, gpt-5.3-codex, gpt-5.4, gpt-5.4-mini, gpt-5.5, gpt-5.6-luna, gpt-5.6-luna-utility, gpt-5.6-sol, gpt-5.6-terra, gpt-6-astra, gpt-6-luna, gpt-6-sol, gpt-6.1-sol, grok-4.5, grok-4.6, grok-4.7, mai-code-1.1-flash, trajectory-compaction, text-embedding-*, plus deprecated gpt-3.5/4/4o/4.1 variants (ignored).
Notes
- Suggested action: add
claude-haiku-5.5togithub-copilotin both models.json files and fix the cache_read values above (runmake refresh-models-json, then adjust). - Deprecated IDs (gpt-4o-mini, gpt-4.1, gpt-4o, gpt-5.4-nano) were ignored per policy.
gpt-5.6-luna-utilityandtrajectory-compactionappear in reflect only (not in SDK); no pricing available.
Generated by 📦 Daily Model Inventory Checker · copilot · auto · 23.6 AIC · ⌖ 0.743 AIC · ⊞ 13.6K · ◷
- expires on Oct 14, 2026, 4:01 PM UTC-08:00
- 主要语言
- Go
- 星标
- 5.4k
- 派生
- 576
- 平均合并
- 8 小时 29 分钟
- 30 天内合并 PR
- 783
环境准备
在浏览器里用你自己的 GitHub 账号启动这个项目的开发容器。
- 提供 Dockerfile 或 Docker Compose 文件
- 没有 Pull Request 模板
- 阅读贡献指南
从这里开始
- 先读完整个 Issue，再读项目的贡献指南。
- 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
- Fork 仓库，在一个分支上完成修改。
- 提交 Pull Request，并在描述里引用这个 Issue 编号。
github/gh-aw 的其他 Issue
-
[duplicate-code] Duplicate Code: pull_request event detection duplicated across safe_update files未关闭automated-analysis code-quality cookie refactoring
难度 2/5 1-3 小时 新手友好度 68/100
维护者通常 1 天内回复
-
[deep-report] Migrate manual os.Setenv/Unsetenv restore patterns to t.Setenv in 2 pkg/cli test files未关闭automation code-quality cookie deep-report improvement quick-win task-mining
难度 2/5 1 小时以内 新手友好度 85/100
维护者通常 1 天内回复
-
automation code-quality cookie deep-report improvement quick-win task-mining
难度 2/5 1 小时以内 新手友好度 82/100
维护者通常 1 天内回复
-
automation code-quality cookie deep-report improvement quick-win task-mining
难度 2/5 1 小时以内 新手友好度 76/100
维护者通常 1 天内回复
-
ai-generated cookie high-priority security
难度 2/5 1-3 小时 新手友好度 68/100
维护者通常 1 天内回复
相似的 Issue
-
难度 1/5 1 小时以内 新手友好度 82/100
维护者通常 1 天内回复
-
enhancement pkg:sdk
难度 2/5 1-3 小时 新手友好度 80/100
aws/aws-durable-execution-sdk-go#144 ·
维护者通常 1 天内回复
-
难度 2/5 1-3 小时 新手友好度 78/100
cloudflare/cloudflared#1761 ·
-
难度 2/5 1-3 小时 新手友好度 72/100
维护者通常 1 天内回复
-
难度 2/5 1-3 小时 新手友好度 72/100
维护者通常 1 天内回复