Skip to content
Source
Sign inSign up
TanStack/schemas

Groq cached-input prices; ledger unpriced Groq and Mistral rows #419

Mergedtombeckenhammerged 2 commits intomain fromrefs/pull/419/headmerged by tombeckenham View on GitHub

tombeckenhamcommented

Part of #411.

Groq

  • Groq's models table has no cached-input column. The prompt-caching doc (https://console.groq.com/docs/prompt-caching.md) lists the models that cache (gpt-oss-20b, gpt-oss-120b, gpt-oss-safeguard-20b). Each of their model pages states its own Cached Input price, which groq-pricing.ts now reads as cache_read_tokens: $0.075/M (120b), $0.037/M (20b, safeguard-20b). The model-page figure is read as stated, not computed from the doc's "50% discount" (20b's page says $0.037, not $0.0375).
  • Fails closed: an empty Supported Models table throws, and so does a listed caching model whose page states no Cached Input price.

Not published (ledgered)

  • groq/allam-2-7b priced + cacheRead: not in the models table; its model page states Input/Output as "Pending".
  • groq/qwen/qwen3.8-27b, meta-llama/llama-prompt-guard-2-22m/86m cacheRead: not in the prompt-caching Supported Models table, no Cached Input on their pages.
  • mistral/labs-leanstral-1-5, labs-leanstral-1-5-1 priced + cacheRead: no row in the pricing table and no price widget on the Leanstral 1.5 page (deprecated 2026-09-29).

All other Mistral chat rows (33/35) are already priced in prod; the snapshot's lower % predates recent parser fixes.

    1. Deploying with  Cloudflare Workers  Cloudflare Workers

      The latest updates on your project. Learn more about integrating Git with Workers.

      StatusNameLatest CommitPreview URLUpdated (UTC)
      ✅ Deployment successful!
      View logs
      modelschemas41bee347Commit Preview URL

      Branch Preview URL
      Oct 22 2026, 11:19 AM
    2. tombeckenham merged thisae93392

    3. tombeckenham closed this

Sign in to join this conversation.