[BOT ISSUE] Cerebras: add missing gemma-4-31b model

[github-actions]

Summary

Google's Gemma 4 31B model is available on the Cerebras public inference platform but is missing from the catalog. Cerebras lists it on both the model overview page and the dedicated model page with full pricing and specs.

The catalog already has two other Cerebras models (gpt-oss-120b and zai-glm-4.7), so the provider is well-established. A separate catalog entry google/gemma-4-31B-it exists for Together AI, but the Cerebras model ID (gemma-4-31b) is different and requires its own entry.

Missing Model

Model ID	Display Name	Input $/1M	Output $/1M	Context	Max Output
gemma-4-31b	Gemma 4 31B	$0.99	$1.49	131,072	40,960

Verification

Check	Status	Detail
Cross-source (model existence)	✅	Listed on Cerebras model overview AND dedicated model page
Cross-source (pricing)	✅	$0.99/$1.49 per 1M tokens confirmed on dedicated model page
Already in catalog	✅	Confirmed gemma-4-31b is not present in model_list.json (searched for gemma-4-31b, gemma.4)
ID format	✅	Matches Cerebras naming convention used by existing entries (gpt-oss-120b, zai-glm-4.7)
Duplicate check	✅	No open issue covers Cerebras gemma-4-31b. Issue #453 covers other Cerebras models but not this one. Issue #761 covers Fireworks Gemma 4 models (different provider and model IDs)

Verification Notes

Field	Source	Notes
Model ID gemma-4-31b	Model overview	Listed alongside gpt-oss-120b and zai-glm-4.7
Pricing $0.99/$1.49	Model detail page	Input and output per 1M tokens
Context 131,072 tokens	Model detail page	Paid tier limit; free tier is 65k
Max output 40,960 tokens	Model detail page	Paid tier limit; free tier is 32k
Multimodal (image input)	Model detail page	Base64 PNG/JPEG, max 5 images, 10MB total
Reasoning support	Model detail page	reasoning_effort parameter supported
displayName not published	—	Derived from model family name; not an official display name

Local Files Inspected

• packages/proxy/schema/model_list.json — confirmed gemma-4-31b is absent; existing Cerebras entries at gpt-oss-120b (line ~1300) and zai-glm-4.7 (line ~9543)
• packages/proxy/schema/models.ts — cerebras is a supported ModelEndpointType

{
  "kind": "missing_model",
  "provider": "cerebras",
  "models": ["gemma-4-31b"],
  "status": "active",
  "model_specs": {
    "gemma-4-31b": {
      "format": "openai",
      "flavor": "chat",
      "multimodal": true,
      "input_cost_per_mil_tokens": 0.99,
      "output_cost_per_mil_tokens": 1.49,
      "displayName": "Gemma 4 31B",
      "reasoning": true,
      "max_input_tokens": 131072,
      "max_output_tokens": 40960,
      "available_providers": ["cerebras"]
    }
  },
  "source_urls": [
    "https://inference-docs.cerebras.ai/models/overview",
    "https://inference-docs.cerebras.ai/models/gemma-4-31b"
  ]
}

[github-actions]

Prepared catalog updates for gemma-4-31b. Added models: gemma-4-31b.

Verification sources: https://inference-docs.cerebras.ai/models/overview, https://inference-docs.cerebras.ai/models/gemma-4-31b
Verified metadata applied:

• gemma-4-31b: format=openai, flavor=chat, displayName=Gemma 4 31B, max_input_tokens=131072, max_output_tokens=40960, pricing(input/output per 1M)=0.99/1.49, providers=cerebras, reasoning=true, multimodal=true

Included in the daily batch PR: #886

[github-actions]

Prepared catalog updates for gemma-4-31b. Added models: gemma-4-31b.

Verification sources: https://inference-docs.cerebras.ai/models/overview, https://inference-docs.cerebras.ai/models/gemma-4-31b
Verified metadata applied:

• gemma-4-31b: format=openai, flavor=chat, displayName=Gemma 4 31B, max_input_tokens=131072, max_output_tokens=40960, pricing(input/output per 1M)=0.99/1.49, providers=cerebras, reasoning=true, multimodal=true

Included in the daily batch PR: #887