[BOT ISSUE] OpenAI: update gpt-5.4-mini and gpt-5.4-nano max_input_tokens from 272000 to 400000

[github-actions]

Summary

The gpt-5.4-mini and gpt-5.4-nano model families have max_input_tokens: 272000 in the catalog, but OpenAI's official documentation states the context window is 400,000 tokens. This affects both the base aliases and their dated snapshots.

The full-size gpt-5.4 correctly has max_input_tokens: 1050000, matching its docs. Only the mini and nano variants are wrong.

Affected Entries

Model ID	Current max_input_tokens	Correct Value	Line
gpt-5.4-mini	272,000	400,000	~158
gpt-5.4-mini-2026-03-17	272,000	400,000	~71
gpt-5.4-nano	272,000	400,000	~174

Also affected (snapshot with the same gap):

• gpt-5.4-nano-2026-03-17 (line ~88) — 272,000 → 400,000

All other fields (pricing, format, reasoning, max_output_tokens) are correct and unchanged.

Verification

Check	Status	Detail
Cross-source (max_input_tokens)	✅	GPT-5.4 mini model page: "400,000 context window" AND OpenAI models overview: comparison table lists GPT-5.4 mini with "Context window: 400K"
Cross-source (nano)	✅	GPT-5.4 nano model page: "400,000 context window"
Already fixed	✅	Confirmed all 4 entries still have max_input_tokens: 272000 in model_list.json
Sibling consistency	✅	gpt-5.4 and gpt-5.4-pro correctly have max_input_tokens: 1050000 — only mini/nano are wrong
Duplicate check	✅	No open issue covers GPT-5.4-mini/nano context window

Verification Notes

Field	Source	Notes
Context window = 400K (mini)	Model page	"400,000 context window"
Context window = 400K (mini)	Models overview	Comparison table confirms 400K
Context window = 400K (nano)	Model page	"400,000 context window"
Max output = 128K	Both model pages	Matches existing max_output_tokens: 128000 — no change needed

Local Files Inspected

• packages/proxy/schema/model_list.json — all 4 entries confirmed at 272000

{
  "kind": "stale_metadata",
  "provider": "openai",
  "models": ["gpt-5.4-mini", "gpt-5.4-mini-2026-03-17", "gpt-5.4-nano"],
  "status": "active",
  "model_specs": {
    "gpt-5.4-mini": {
      "format": "openai",
      "flavor": "chat",
      "multimodal": true,
      "input_cost_per_mil_tokens": 0.75,
      "output_cost_per_mil_tokens": 4.5,
      "input_cache_read_cost_per_mil_tokens": 0.075,
      "displayName": "GPT-5.4 mini",
      "reasoning": true,
      "max_input_tokens": 400000,
      "max_output_tokens": 128000,
      "available_providers": ["openai", "azure"]
    },
    "gpt-5.4-mini-2026-03-17": {
      "format": "openai",
      "flavor": "chat",
      "multimodal": true,
      "input_cost_per_mil_tokens": 0.75,
      "output_cost_per_mil_tokens": 4.5,
      "input_cache_read_cost_per_mil_tokens": 0.075,
      "displayName": "GPT-5.4 mini (2026-03-17)",
      "reasoning": true,
      "parent": "gpt-5.4-mini",
      "max_input_tokens": 400000,
      "max_output_tokens": 128000,
      "available_providers": ["openai", "azure"]
    },
    "gpt-5.4-nano": {
      "format": "openai",
      "flavor": "chat",
      "multimodal": true,
      "input_cost_per_mil_tokens": 0.2,
      "output_cost_per_mil_tokens": 1.25,
      "input_cache_read_cost_per_mil_tokens": 0.02,
      "displayName": "GPT-5.4 nano",
      "reasoning": true,
      "max_input_tokens": 400000,
      "max_output_tokens": 128000,
      "available_providers": ["openai", "azure"]
    }
  },
  "source_urls": [
    "https://developers.openai.com/api/docs/models/gpt-5.4-mini",
    "https://developers.openai.com/api/docs/models/gpt-5.4-nano",
    "https://developers.openai.com/api/docs/models"
  ]
}

[github-actions]

Prepared catalog updates for gpt-5.4-mini, gpt-5.4-mini-2026-03-17, gpt-5.4-nano. Updated models: gpt-5.4-mini, gpt-5.4-mini-2026-03-17, gpt-5.4-nano.

Verification sources: https://developers.openai.com/api/docs/models/gpt-5.4-mini, https://developers.openai.com/api/docs/models/gpt-5.4-nano, https://developers.openai.com/api/docs/models
Verified metadata applied:

• gpt-5.4-mini: format=openai, flavor=chat, displayName=GPT-5.4 mini, max_input_tokens=400000, max_output_tokens=128000, pricing(input/output per 1M)=0.75/4.5, input_cache_read_cost_per_mil_tokens=0.075, providers=openai|azure, reasoning=true, multimodal=true
• gpt-5.4-mini-2026-03-17: format=openai, flavor=chat, displayName=GPT-5.4 mini (2026-03-17), parent=gpt-5.4-mini, max_input_tokens=400000, max_output_tokens=128000, pricing(input/output per 1M)=0.75/4.5, input_cache_read_cost_per_mil_tokens=0.075, providers=openai|azure, reasoning=true, multimodal=true
• gpt-5.4-nano: format=openai, flavor=chat, displayName=GPT-5.4 nano, max_input_tokens=400000, max_output_tokens=128000, pricing(input/output per 1M)=0.2/1.25, input_cache_read_cost_per_mil_tokens=0.02, providers=openai|azure, reasoning=true, multimodal=true

Included in the daily batch PR: #904