[BOT ISSUE] Cerebras: remove stale provider mappings for llama3.1-8b, llama3.3-70b, and llama-4-scout-17b-16e-instruct

[github-actions]

Summary

Cerebras Inference has reduced its shared inference catalog to only 2 active models: gpt-oss-120b (production) and zai-glm-4.7 (preview). The proxy catalog still maps several discontinued models to Cerebras as their sole provider, which would cause routing failures for users requesting these models.

Affected Models

Model ID	Current Provider Mapping	Status on Cerebras
llama3.1-8b	cerebras (only provider)	Deprecated May 27, 2026
llama3.3-70b	cerebras (only provider)	No longer available
llama-4-scout-17b-16e-instruct	cerebras (only provider)	No longer available

Suggested Fix

These model IDs should either:

1. Be marked as deprecated: true in model_list.json (since their only provider no longer serves them), OR
2. Have their cerebras provider mapping removed from index.ts and the entries cleaned up

Note: deepseek-r1-distill-llama-70b also has a stale cerebras mapping in index.ts (line ~412), but retains groq as a secondary provider — though Groq also deprecated it in October 2025, leaving it with no active providers either.

Sources

• Cerebras Models Overview — only lists gpt-oss-120b and zai-glm-4.7 as currently available models
• AI Inference Providers Compared (2026) — Infrabase — confirms Cerebras deprecated Llama 3.1 8B and Qwen 3 235B Instruct on May 27, 2026, reducing active catalog to 2 models

Existing Catalog References

• llama3.1-8b: model_list.json line ~6110 (available_providers: ["cerebras"]), index.ts line ~448
• llama3.3-70b: model_list.json line ~5721, index.ts line ~449
• llama-4-scout-17b-16e-instruct: model_list.json line ~5712, index.ts line ~447

[github-actions]

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

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

• 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=cerebras, deprecated=true
• 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
• 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

Included in the daily batch PR: #790