[model-inventory] Model alias inventory update - 2026-07-10

[github-actions]

Summary

Daily model inventory scan for 2026-07-10. All existing Copilot model alias patterns have adequate coverage. Three new GPT-5.6 sub-models (gpt-5.6-luna, gpt-5.6-sol, gpt-5.6-terra) are live in the Copilot reflect endpoint but missing pricing entries in models.json. Billing docs also reveal a new "Claude Opus 4.8 (fast mode)" variant and confirm pricing for existing entries.

• Providers queried: OpenAI (219), Anthropic (9), Gemini (50), Copilot/reflect (35)
• Total models found: 313 across all providers
• Proposed alias changes: 0 (all live models already covered)
• Pricing gaps found: 4 (3 missing, 1 new variant)

Provider Model Counts

Provider	Models Available	Status
openai	219	✅ ok
anthropic	9	✅ ok
gemini	50	✅ ok
copilot	35	✅ ok (reflect endpoint)

Raw API Fields Discovered

Anthropic (/raw.json → .data[]):

• id, display_name, created_at — model identity
• max_input_tokens (up to 1M), max_tokens (output limit, up to 128K)
• capabilities.thinking.supported — extended reasoning flag
• capabilities.image_input.supported — vision flag
• capabilities.batch.supported, capabilities.citations.supported, capabilities.code_execution.supported
• capabilities.effort.{low,medium,high,xhigh,max} — effort-level support

OpenAI (/raw.json → .data[]):

• id, object, created (unix timestamp), owned_by — minimal public fields
• No pricing, capability, or context-window metadata exposed in list endpoint

Gemini (/raw.json → .models[]):

• name, displayName, description, version
• inputTokenLimit, outputTokenLimit — useful for tier inference
• supportedGenerationMethods — capability flags (generateContent, batchGenerateContent, etc.)
• temperature, topP, topK, maxTemperature — sampling parameters
• thinking: true — extended reasoning flag (on flash and pro variants)

Copilot (reflect endpoint):

• Model IDs as a flat string array (no per-model metadata in current reflect response)
• Billing multipliers and vendor metadata not returned in this run (models returned as strings, not objects)

models.json Pricing Analysis

Missing from models.json (github-copilot provider)

These models are live in the Copilot reflect endpoint and confirmed in the GitHub Docs billing table, but have no pricing entry in pkg/cli/data/models.json:

Model ID	Provider	Inferred Category	Docs Pricing (input / output per 1M)	Basis
gpt-5.6-luna	copilot	Lightweight	$1.00 / $6.00 (default ≤200K)	Billing docs + reflect
gpt-5.6-sol	copilot	Powerful	$5.00 / $30.00 (default ≤272K)	Billing docs + reflect
gpt-5.6-terra	copilot	Versatile	$2.50 / $15.00 (default ≤272K)	Billing docs + reflect

Note: Billing docs also list "Claude Opus 4.8 (fast mode) (preview)" at $10.00 / $50.00 input/output per 1M tokens — a distinct model variant from claude-opus-4.8. It is not yet in the reflect endpoint model list; treat as pending until a stable model ID is published.

Proposed additions to pkg/cli/data/models.json (and mirror actions/setup/js/models.json):

"gpt-5.6-luna": {
  "cost": {
    "input": "1e-06",
    "output": "6e-06",
    "cache_read": "1e-07"
  },
  "provider_type": "openai",
  "wire_api": "responses"
},
"gpt-5.6-sol": {
  "cost": {
    "input": "5e-06",
    "output": "3e-05",
    "cache_read": "5e-07"
  },
  "provider_type": "openai",
  "wire_api": "responses"
},
"gpt-5.6-terra": {
  "cost": {
    "input": "2.5e-06",
    "output": "1.5e-05",
    "cache_read": "2.5e-07"
  },
  "provider_type": "openai",
  "wire_api": "responses"
}

Historical entries not currently returned

These models appear in models.json but are absent from the Copilot reflect endpoint. They should be retained as historical records:

Model ID	Last Known Provider	Note
claude-fable-5	github-copilot	In billing docs; may not be reflect-accessible yet
claude-sonnet-4	github-copilot	Older generation; historical
gemini-3.1-pro-preview	github-copilot	Preview phase; still in billing docs
gemini-3.5-flash	github-copilot	Not in reflect; still in billing docs
gpt-5.2	github-copilot	Superseded by 5.3/5.4; historical
gpt-5.2-codex	github-copilot	Superseded; historical
gpt-5.4-nano	github-copilot	Intentionally deprecated/excluded; retain
kimi-k2.7-code	github-copilot	In billing docs but not reflect; historical
raptor-mini	github-copilot	In billing docs but not reflect; historical

Inferred vs stored pricing discrepancies

No confirmed multiplier discrepancies found. The reflect endpoint returned model IDs as a flat string array (no billing.multiplier per-model metadata in this run), so multiplier-level validation was not possible. Billing docs pricing is consistent with existing models.json cost entries for all models present in both.

Model ID	Stored Input (per tok)	Docs Input (per 1M)	Match
gpt-5-mini	2.5e-7	$0.25	✅
gpt-5.3-codex	1.75e-6	$1.75	✅
gpt-5.4	2.5e-6	$2.50	✅
gpt-5.4-mini	7.5e-7	$0.75	✅
gpt-5.5	5e-6	$5.00	✅
claude-haiku-4.5	1e-6	$1.00	✅
claude-sonnet-4.5	3e-6	$3.00	✅
claude-sonnet-4.6	3e-6	$3.00	✅
claude-opus-4.5	5e-6	$5.00	✅
claude-opus-4.6	5e-6	$5.00	✅
claude-opus-4.7	5e-6	$5.00	✅
claude-opus-4.8	5e-6	$5.00	✅
claude-sonnet-5	2e-6	$2.00	✅
claude-fable-5	1e-5	$10.00	✅
gemini-2.5-pro	1.25e-6	$1.25	✅
gemini-3-flash-preview	5e-7	$0.50	✅
gemini-3.5-flash	1.5e-6	$1.50	✅
raptor-mini	2.5e-7	$0.25	✅
kimi-k2.7-code	9.5e-7	$0.95	✅
mai-code-1-flash-picker	7.5e-7	$0.75	✅

Proposed Alias Updates

No alias changes are required. All 35 live Copilot models match at least one existing alias pattern:

• gpt-5.6-luna, gpt-5.6-sol, gpt-5.6-terra → covered by "gpt-5.6": ["copilot/gpt-5.6*", "openai/gpt-5.6*"]
• All claude-opus-4.* variants → covered by "opus": ["copilot/*opus*", "anthropic/*opus*"]
• All claude-sonnet-4.* and claude-sonnet-5 → covered by "sonnet": ["copilot/*sonnet*", ...]
• claude-haiku-4.5 → covered by "haiku": ["copilot/*haiku*", ...]
• gemini-3-flash-preview → covered by "gemini-flash" and "gemini-3-flash" aliases
• gemini-2.5-pro → covered by "gemini-pro": ["copilot/gemini-*pro*", ...]
• mai-code-1-flash-picker → covered by "mai-code": ["copilot/mai-code*", ...]
• text-embedding-* models → embedding models, no alias needed (not LLM targets)

Full Model Lists by Provider

Copilot (reflect, 35 models)

claude-haiku-4.5, claude-opus-4.5, claude-opus-4.6, claude-opus-4.7, claude-opus-4.8,
claude-sonnet-4.5, claude-sonnet-4.6, claude-sonnet-5, gemini-2.5-pro,
gemini-3-flash-preview, gpt-3.5-turbo, gpt-3.5-turbo-0613, gpt-4, gpt-4-0613,
gpt-4-o-preview, gpt-4.1, gpt-4.1-2025-04-14, gpt-4o, gpt-4o-2024-05-13,
gpt-4o-2024-08-06, gpt-4o-2024-11-20, gpt-4o-mini, gpt-4o-mini-2024-07-18,
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, mai-code-1-flash-picker, text-embedding-3-small,
text-embedding-3-small-inference, text-embedding-ada-002

Anthropic API (9 models)

claude-haiku-4-5-20251001, claude-opus-4-1-20250805, claude-opus-4-5-20251101,
claude-opus-4-6, claude-opus-4-7, claude-opus-4-8, claude-sonnet-4-5-20250929,
claude-sonnet-4-6, claude-sonnet-5

Gemini API (50 models, selection of text models)

antigravity-preview-05-2026, gemini-2.0-flash, gemini-2.0-flash-001, gemini-2.0-flash-lite,
gemini-2.5-flash, gemini-2.5-flash-lite, gemini-2.5-pro, gemini-3-flash-preview,
gemini-3-pro-preview, gemini-3.1-flash-lite, gemini-3.1-flash-lite-preview,
gemini-3.1-pro-preview, gemini-3.5-flash, gemma-4-26b-a4b-it, gemma-4-31b-it,
nano-banana-pro-preview, (+ TTS, audio, image, robotics, embedding models)

OpenAI API (219 models, notable)

gpt-4.1, gpt-4.1-mini, gpt-4.1-nano, gpt-4o-mini-2024-07-18, gpt-5, gpt-5-codex,
gpt-5-mini, gpt-5-nano, gpt-5-pro, gpt-5.1-codex, gpt-5.2-codex, gpt-5.3-codex,
gpt-5.4, gpt-5.4-mini, gpt-5.4-nano, gpt-5.5, gpt-5.6-luna, gpt-5.6-sol, gpt-5.6-terra,
o1, o1-pro, o3, o3-mini, o3-pro, o4-mini (+ many research/realtime/audio/image variants)

Notes

• gpt-4o-mini, gpt-4.1, gpt-4o, and gpt-5.4-nano are intentionally deprecated Copilot-facing model IDs and excluded from missing/discrepancy analysis per standing policy.
• The "Claude Sonnet 51" entry in billing docs is a likely rendering artifact for claude-sonnet-5 (i.e. "5" rendered as "51" in the page scrape); treated as claude-sonnet-5.
• "Claude Opus 4.8 (fast mode) (preview)" appears in billing docs at $10/$50 per 1M tokens but has no stable model ID yet — monitor for a canonical ID in future reflect snapshots.
• The Copilot reflect endpoint returned model IDs as a plain string array (not enriched objects), so per-model billing.multiplier validation was skipped this run.
• gemini-3.1-pro-preview is listed in billing docs (at $2.00/$12.00 default, $4.00/$18.00 long-context) but not returned by the reflect endpoint — keep as historical entry.

Warning

Firewall blocked 7 domains

The following domains were blocked by the firewall during workflow execution:

• accounts.google.com
• android.clients.google.com
• clients2.google.com
• collector.githubapp.com
• contentautofill.googleapis.com
• safebrowsingohttpgateway.googleapis.com
• www.google.com

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

network:
  allowed:
    - defaults
    - "accounts.google.com"
    - "android.clients.google.com"
    - "clients2.google.com"
    - "collector.githubapp.com"
    - "contentautofill.googleapis.com"
    - "safebrowsingohttpgateway.googleapis.com"
    - "www.google.com"

See Network Configuration for more information.

Generated by 📦 Daily Model Inventory Checker · 97.6 AIC · ⌖ 12.2 AIC · ⊞ 8.5K · ◷

• expires on Jul 16, 2026, 4:28 PM UTC-08:00