[BOT ISSUE] Cerebras: remove deprecated models (`llama3.3-70b`, `llama3.1-8b`, `llama-4-scout-17b-16e-instruct`)

[github-actions]

Summary

Cerebras Inference has deprecated all models except gpt-oss-120b and zai-glm-4.7. The catalog still lists four other models with cerebras as a provider. Three of them (llama3.3-70b, llama3.1-8b, llama-4-scout-17b-16e-instruct) have cerebras as their only provider and should be deprecated or removed entirely. A fourth (deepseek-r1-distill-llama-70b) has cerebras alongside groq and should only have cerebras removed from its provider list.

Affected Models

Model ID	Cerebras Deprecation Date	Other Providers	Action
llama3.3-70b	2026-02-16	None	Deprecate model or remove entry
llama3.1-8b	2026-05-27	None	Deprecate model or remove entry
llama-4-scout-17b-16e-instruct	2025-11-03	None	Deprecate model or remove entry
deepseek-r1-distill-llama-70b	2025-08-12	groq	Remove cerebras from providers

All four models appear on the Cerebras Deprecation page with past deprecation dates. The Cerebras Models Overview now lists only two production/preview models: gpt-oss-120b and zai-glm-4.7.

Evidence

1. Cerebras Models Overview (https://inference-docs.cerebras.ai/models/overview): Lists only gpt-oss-120b (production) and zai-glm-4.7 (preview) as available models.
2. Cerebras Deprecation Page (https://inference-docs.cerebras.ai/support/deprecation): Explicitly lists deprecation dates for each removed model, with suggested replacements pointing to gpt-oss-120b.

Current Catalog Entries

llama3.3-70b (line ~6103 in model_list.json; AvailableEndpointTypes in index.ts maps to ["cerebras"]):

"llama3.3-70b": {
  "format": "openai",
  "flavor": "chat",
  "input_cost_per_mil_tokens": 0.1,
  "output_cost_per_mil_tokens": 0.1,
  "displayName": "Llama 3.3 70B"
}

llama3.1-8b (line ~6110; AvailableEndpointTypes maps to ["cerebras"]):

"llama3.1-8b": {
  "format": "openai",
  "flavor": "chat",
  "input_cost_per_mil_tokens": 0.1,
  "output_cost_per_mil_tokens": 0.1,
  "displayName": "Llama 3.1 8B",
  "max_input_tokens": 128000,
  "max_output_tokens": 128000
}

llama-4-scout-17b-16e-instruct (line ~5712; AvailableEndpointTypes maps to ["cerebras"]):

"llama-4-scout-17b-16e-instruct": {
  "format": "openai",
  "flavor": "chat",
  "input_cost_per_mil_tokens": 0.11,
  "output_cost_per_mil_tokens": 0.34,
  "displayName": "LLaMA 4 Scout 17B 16e Instruct",
  "max_input_tokens": 131072,
  "max_output_tokens": 8192
}

deepseek-r1-distill-llama-70b (line ~3536; AvailableEndpointTypes maps to ["groq", "cerebras"]):

"deepseek-r1-distill-llama-70b": {
  "format": "openai",
  "flavor": "chat",
  "input_cost_per_mil_tokens": 0.23,
  "output_cost_per_mil_tokens": 0.69,
  "displayName": "DeepSeek R1 Distill Llama 70b",
  "experimental": true
}

Verification Checklist

Check	Status	Detail
Cross-source	✅	All four models confirmed absent from Cerebras Models Overview AND explicitly listed on Cerebras Deprecation page with past deprecation dates
Recent commits	✅	No commit in the past 30 days touches Cerebras provider mappings for these models
Duplicate check	✅	No existing open issue covers Cerebras deprecations

Local Files Inspected

• packages/proxy/schema/model_list.json — confirmed all four models still have entries
• packages/proxy/schema/index.ts — confirmed AvailableEndpointTypes still maps these models to cerebras

{
  "kind": "deprecated_models",
  "provider": "cerebras",
  "models": ["llama3.3-70b", "llama3.1-8b", "llama-4-scout-17b-16e-instruct"],
  "status": "deprecated",
  "source_urls": [
    "https://inference-docs.cerebras.ai/models/overview",
    "https://inference-docs.cerebras.ai/support/deprecation"
  ]
}

[github-actions]

Prepared catalog updates for llama3.3-70b, llama3.1-8b, llama-4-scout-17b-16e-instruct. Updated models: llama3.3-70b, llama3.1-8b, llama-4-scout-17b-16e-instruct.

Verification sources: https://inference-docs.cerebras.ai/models/overview, https://inference-docs.cerebras.ai/support/deprecation
Verified metadata applied:

• llama3.3-70b: format=openai, flavor=chat, displayName=Llama 3.3 70B, max_input_tokens=not provided, max_output_tokens=not provided, pricing(input/output per 1M)=0.1/0.1, providers=not provided, deprecated=true, deprecation_date=2026-02-16
• llama3.1-8b: format=openai, flavor=chat, displayName=Llama 3.1 8B, max_input_tokens=128000, max_output_tokens=128000, pricing(input/output per 1M)=0.1/0.1, providers=not provided, deprecated=true, deprecation_date=2026-05-27
• llama-4-scout-17b-16e-instruct: format=openai, flavor=chat, displayName=LLaMA 4 Scout 17B 16e Instruct, max_input_tokens=131072, max_output_tokens=8192, pricing(input/output per 1M)=0.11/0.34, providers=not provided, deprecated=true, deprecation_date=2025-11-03

Included in the daily batch PR: #822