[BOT ISSUE] Mistral: codestral-2508 stale max_input_tokens (256000 → 128000) and voxtral-small-2507 stale output pricing ($0.4 → $0.3)

[github-actions]

Summary

Two Mistral-only models have stale metadata in the catalog:

1. codestral-2508 has max_input_tokens: 256000, but Mistral's model card documents a 128k context window. The parent alias codestral-latest correctly has max_input_tokens: 128000.

2. voxtral-small-2507 has output_cost_per_mil_tokens: 0.4, but Mistral's model card documents $0.3/M output tokens.

Model ID	Field	Catalog	Mistral
codestral-2508	max_input_tokens	256,000	128,000
voxtral-small-2507	output_cost_per_mil_tokens	0.4	0.3

Sources

• Codestral 25.08 model card — states "128k" context and pricing of $0.3/$0.9 per M tokens
• Voxtral Small model card — states output pricing of "$0.3/M Tokens"
• codestral-latest entry in model_list.json — correctly has max_input_tokens: 128000, confirming the snapshot is wrong

Verification

Check	Status	Detail
Cross-source (codestral)	✅	128k confirmed on model card page AND by parent alias codestral-latest having 128000 in catalog
Cross-source (voxtral)	✅	$0.3 output confirmed on model card page; web search also returns $0.3
Already fixed	✅	Confirmed codestral-2508 has 256000 at line ~5202 and voxtral-small-2507 has 0.4 at line ~13343
ID format	✅	Both IDs match existing catalog entries
Duplicate check	✅	No open issue covers these gaps

Verification Notes

Field	Source	Notes
max_input_tokens: 128000 (codestral)	Codestral 25.08 model card	Model card explicitly states "128k" context
max_input_tokens: 128000 (codestral)	codestral-latest in catalog	Parent alias already has the correct value
output_cost_per_mil_tokens: 0.3 (voxtral)	Voxtral Small model card	States "$0.3/M Tokens" for output
Codestral pricing correct	Model card	$0.3/$0.9 matches catalog
Voxtral input pricing correct	Model card	$0.1 matches catalog

Local Files Inspected

• packages/proxy/schema/model_list.json — codestral-2508 at line ~5202 with max_input_tokens: 256000; voxtral-small-2507 at line ~13343 with output_cost_per_mil_tokens: 0.4

{
  "kind": "stale_metadata",
  "provider": "mistral",
  "models": ["codestral-2508", "voxtral-small-2507"],
  "status": "active",
  "model_specs": {
    "codestral-2508": {
      "format": "openai",
      "flavor": "chat",
      "input_cost_per_mil_tokens": 0.3,
      "output_cost_per_mil_tokens": 0.9,
      "displayName": "Codestral 2508",
      "parent": "codestral-latest",
      "max_input_tokens": 128000,
      "available_providers": ["mistral"]
    },
    "voxtral-small-2507": {
      "format": "openai",
      "flavor": "chat",
      "multimodal": true,
      "input_cost_per_mil_tokens": 0.1,
      "output_cost_per_mil_tokens": 0.3,
      "max_input_tokens": 32000,
      "available_providers": ["mistral"]
    }
  },
  "source_urls": [
    "https://docs.mistral.ai/models/model-cards/codestral-25-08",
    "https://docs.mistral.ai/models/voxtral-small-25-07"
  ]
}

[github-actions]

Prepared catalog updates for codestral-2508, voxtral-small-2507. Updated models: codestral-2508, voxtral-small-2507.

Verification sources: https://docs.mistral.ai/models/model-cards/codestral-25-08, https://docs.mistral.ai/models/voxtral-small-25-07
Verified metadata applied:

• codestral-2508: format=openai, flavor=chat, displayName=Codestral 2508, parent=codestral-latest, max_input_tokens=128000, max_output_tokens=not provided, pricing(input/output per 1M)=0.3/0.9, providers=mistral
• voxtral-small-2507: format=openai, flavor=chat, max_input_tokens=32000, max_output_tokens=not provided, pricing(input/output per 1M)=0.1/0.3, providers=mistral, multimodal=true

Included in the daily batch PR: #876