[model-inventory] Model alias inventory update - 2026-09-26

[github-actions]

Summary

Ran the full inventory refresh (OpenAI, Anthropic, Gemini, Copilot SDK, Copilot reflect, docs pricing table, models.dev catalog). All 5 live-model sources are fully covered by the existing model_aliases.json glob patterns — no new alias generations are unmapped. However, pkg/cli/data/models.json (and its mirror actions/setup/js/models.json) was stale relative to the models.dev catalog and has been refreshed, with all previously-tracked historical entries preserved.

• Providers queried: OpenAI, Anthropic, Gemini, Copilot SDK, Copilot (reflect), models.dev catalog, GitHub Docs pricing table
• Total live models found: 134 (OpenAI) + 10 (Anthropic) + 50 (Gemini) + 22 (Copilot SDK) + 40 (Copilot reflect)
• Proposed alias changes: 0 (full coverage confirmed)
• Pricing gaps found: 0 discrepancies; models.json refreshed from models.dev catalog with historical entries restored

Provider Model Counts

Provider	Models Available	Status
openai	134	✅ ok
anthropic	10	✅ ok
gemini	50	✅ ok
copilot-sdk	22	✅ ok
copilot (reflect)	40	✅ ok

Raw API Fields Discovered

• OpenAI: id, owned_by, created — no pricing/capability metadata in the raw /v1/models response; naming conventions (-mini, -nano, o1/o3/o4) used to cross-check tiers.
• Anthropic: id, display_name, created_at, type — no pricing in raw response; family grouping (haiku/sonnet/opus) used for tier validation.
• Gemini: id, display_name, description, input_token_limit, output_token_limit, supported_generation_methods, version — token limits used as a complexity/tier proxy.
• Copilot SDK: rich capabilities.limits (context/output/prompt token limits, vision support with media types), capabilities.supports (reasoning effort, thinking budget, tool calls, streaming, structured outputs), and billing.tokenPrices (input/output/cache prices, long-context tier pricing, contextMax/maxPromptTokens thresholds) — the most complete pricing source available.
• Copilot reflect: pricing.default / pricing.long_context with input/output/cachedInput/cacheWrite/threshold — matches Copilot SDK pricing exactly for every model where both are present.
• Docs pricing table (docs.github.com/.../models-and-pricing): Model, Release status, Category, Tier, Threshold (input tokens), Input, Cached input, Cache write, Output — only lists the OpenAI/GPT family (no Claude/Gemini rows); confirms Copilot SDK and reflect pricing for gpt-5-mini, gpt-5.3-codex, gpt-5.4[-mini], gpt-5.5, gpt-5.6-{luna,sol,terra}, gpt-6-{astra,luna,sol}.
• All of the above are already captured (where relevant) in the enriched models.json artifacts fetched during this run.

models.json Pricing Analysis

Missing from models.json

None. Cross-checking Copilot SDK billing.tokenPrices, Copilot reflect pricing, and the docs pricing table against pkg/cli/data/models.json for every Copilot model with pricing data available produced zero mismatches — all input/output cost fields agree exactly across all three sources.

gpt-4o-mini, gpt-4.1, gpt-4o, and gpt-5.4-nano were excluded from this comparison per the deprecated-model policy, even though they still appear in the Copilot reflect endpoint.

Historical entries not currently returned

models.json was refreshed from the models.dev catalog via make refresh-models-json. The refresh alone would have dropped a number of entries that are no longer present upstream; these have been restored as historical records (not deleted) in both pkg/cli/data/models.json and actions/setup/js/models.json:

• Anthropic: claude-3-5-haiku-20241022, claude-3-5-haiku-latest, claude-3-5-sonnet-20240620, claude-3-5-sonnet-20241022, claude-3-7-sonnet-20250219, claude-3-haiku-20240307, claude-3-opus-20240229, claude-3-sonnet-20240229, claude-fable-5.1, claude-opus-4-0, claude-opus-4-1, claude-opus-4-1-20250805, claude-opus-4-20250514, claude-sonnet-4-0, claude-sonnet-4-20250514
• OpenAI: github/grok-4.5, gpt-5-chat-latest, gpt-5-codex, gpt-5-search-api, gpt-5.1-chat-latest, gpt-5.1-codex, gpt-5.1-codex-max, gpt-5.1-codex-mini, gpt-5.2-codex, o1-mini, o1-preview, o3-deep-research, o4-mini-deep-research
• Copilot: claude-opus-4.5, claude-opus-4.6, claude-sonnet-4, claude-sonnet-4.5, gemini-2.5-pro, gemini-3-flash-preview, gemini-3.1-pro-preview, gpt-4.1, gpt-5.2, gpt-5.2-codex, raptor-mini

New entries picked up by the refresh (all with pricing cross-validated against Copilot SDK/reflect/docs where applicable): claude-opus-5.5, gpt-6-luna, gpt-6-sol, plus the full GPT-5.3/5.4/5.5/5.6/6-family and Gemini 3.5–3.8-flash lines already present.

Inferred vs stored pricing discrepancies

None found. All checked models matched exactly across Copilot SDK, Copilot reflect, and the docs pricing table.

Proposed Alias Updates

No alias changes are proposed this run. Verification method: every live model ID from Copilot SDK, Copilot reflect, Anthropic, Gemini, and OpenAI (excluding non-chat modalities: embeddings, TTS, whisper, audio, realtime, transcribe, image, moderation, sora, legacy completions, search-preview endpoints) was checked against the full recursively-resolved set of glob patterns in pkg/workflow/data/model_aliases.json. Zero unmatched models were found in any provider, including the newest generations (gpt-6-*, gpt-5.6-*, claude-opus-5.5, gemini-3.8-flash, grok-4.7, mai-code-1.1-flash).

Full Model Lists by Provider

Copilot SDK (22): auto, claude-haiku-4.5, claude-opus-4.7, claude-opus-4.8, claude-opus-5, claude-opus-5.5, claude-sonnet-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-sol, gpt-5.6-terra, gpt-6-astra, gpt-6-luna, gpt-6-sol, grok-4.5, grok-4.6, grok-4.7, mai-code-1.1-flash

Copilot reflect (40): claude-haiku-4.5, claude-opus-4.7, claude-opus-4.8, claude-opus-5, claude-opus-5.5, claude-sonnet-5, gpt-3.5-turbo[-0613], gpt-4[-0613], gpt-4-o-preview, gpt-4.1[-2025-04-14], gpt-41-copilot, gpt-4o[-2024-05-13/08-06/11-20], gpt-4o-mini[-2024-07-18], gpt-5-mini, gpt-5.3-codex, gpt-5.4[-mini], gpt-5.5, gpt-5.6-luna[-utility], gpt-5.6-sol, gpt-5.6-terra, gpt-6-astra, gpt-6-luna, gpt-6-sol, grok-4.5/4.6/4.7, mai-code-1.1-flash, text-embedding-3-small[-inference], text-embedding-ada-002, trajectory-compaction

Anthropic (10): claude-haiku-4-5-20251001, claude-opus-4-5-20251101, claude-opus-4-6, claude-opus-4-7, claude-opus-4-8, claude-opus-5, claude-opus-5-5, claude-sonnet-4-5-20250929, claude-sonnet-4-6, claude-sonnet-5

Gemini (50, selected): gemini-2.5-{flash,flash-image,flash-lite,pro}, gemini-3-{flash-preview,pro-image[-preview]}, gemini-3.1-{flash-image[-preview],flash-lite[-image],flash-lite-preview,flash-tts-preview,pro-preview[-customtools]}, gemini-3.5-flash[-lite], gemini-3.6/3.7/3.8-flash[-lite-tts], gemini-{embedding-001/2[-preview]}, gemini-{flash,pro}-latest, gemini-omni-1.1-flash, gemini-robotics-er-2-preview, gemma-4-{26b,31b}*, lyria-3[.5][-pro]-preview, veo-3.1-{fast-,}generate-preview, antigravity-preview-*, deep-research-*, aqa

OpenAI (134, selected new/notable): full gpt-5.x/gpt-6-* family (chat, codex, pro, mini/nano variants), o1/o3/o4 reasoning + -pro/-deep-research variants, gpt-image-*, gpt-audio-*, gpt-realtime-*, sora-2[-pro], embeddings/tts/whisper (non-chat, excluded from alias-coverage check)

Notes

• Copilot pricing validation is fully green: Copilot SDK billing.tokenPrices, Copilot reflect pricing, and the GitHub Docs pricing table all agree exactly on input/output costs for every GA model with pricing data (21 models cross-checked, 0 mismatches).
• The docs pricing table only covers the OpenAI/GPT-family Copilot models; Claude and Gemini Copilot-hosted models have no equivalent public docs table row, so validation for those relies on SDK + reflect agreement only (which also matched exactly).
• gpt-4o-mini, gpt-4.1, gpt-4o, and gpt-5.4-nano remain intentionally excluded/deprecated per policy and were not reintroduced despite appearing in Copilot reflect data.
• models.json was refreshed via make refresh-models-json (source: models.dev/catalog.json); all previously-tracked historical model IDs that the refresh would have dropped were manually restored in both mirror files to avoid silent data loss. pkg/cli/data/models.json and actions/setup/js/models.json remain byte-identical (make validate-models-json-sync passes).
• Alias coverage is comprehensive: sonnet-6x, gpt-6, gpt-5.6, grok, mai-code, and the versioned gemini-3.x-flash aliases already anticipate and match the newest model generations discovered in this run.
• No new task-oriented alias gaps were identified — coding (→ gpt-5*codex*/kimi), reasoning (→ o1*/o3*/o4*), summarization (→ haiku/gpt-5-mini/gemini-flash-lite), and vision (→ gemini-*image*/gemini-*flash*) aliases already cover the semantic use cases called out in the task brief.

Warning

Firewall blocked 1 domain

The following domain was blocked by the firewall during workflow execution:

• clients2.google.com

To allow these domains, add them to the network.allowed list in your workflow frontmatter:

network:
  allowed:
    - defaults
    - "clients2.google.com"

See Network Configuration for more information.

Generated by 📦 Daily Model Inventory Checker · copilot · auto · 138 AIC · ⌖ 11.4 AIC · ⊞ 13.7K · ◷

• expires on Oct 2, 2026, 4:07 PM UTC-08:00

[github-actions]

This issue is being closed as outdated. A newer issue has been created: #63698

View newer issue

---

This action was performed automatically by the Daily Model Inventory Checker workflow.