[BOT ISSUE] Together: fix Qwen3.5-397B-A17B cached input pricing from $0.06 to $0.35

[github-actions]

Provider

Together

Model IDs

• Qwen/Qwen3.5-397B-A17B

Kind

stale_metadata

Current catalog values

{
  "input_cost_per_mil_tokens": 0.6,
  "output_cost_per_mil_tokens": 3.6,
  "input_cache_read_cost_per_mil_tokens": 0.06,
  "available_providers": ["together"]
}

Proposed update

{
  "input_cache_read_cost_per_mil_tokens": 0.35
}

Details

The Qwen/Qwen3.5-397B-A17B entry has input_cache_read_cost_per_mil_tokens: 0.06, but Together AI's official pricing page lists the cached input cost at $0.35 per million tokens. The input ($0.60) and output ($3.60) prices are correct.

The $0.06 value appears to be incorrect — possibly from an older sync or a data source error. This model is only available on Together (no other providers), so Together's pricing page is the authoritative source.

Sources

• Together AI pricing page: https://www.together.ai/pricing — shows Qwen3.5-397B-A17B at $0.60 input ($0.35 cached) / $3.60 output
• Verified twice via independent page fetches

Verification checklist

• Cross-source: Together pricing page confirms $0.35 cached input (verified twice)
• Already-fixed: Catalog still shows $0.06, not $0.35
• ID format: Qwen/Qwen3.5-397B-A17B already exists in model_list.json

[github-actions]

Prepared catalog updates for Qwen/Qwen3.5-397B-A17B. Updated models: Qwen/Qwen3.5-397B-A17B.

Verification sources: https://www.together.ai/pricing
Verified metadata applied:

• Qwen/Qwen3.5-397B-A17B: format=openai, flavor=chat, displayName=Qwen3.5 397B A17B, max_input_tokens=262144, max_output_tokens=not provided, pricing(input/output per 1M)=0.6/3.6, input_cache_read_cost_per_mil_tokens=0.35, providers=together

Included in the daily batch PR: #904