Follow Discord
Sweep 06 Nov 2026 · 01:32Z Build v2.1.319 537 read Stable v2.1.311 Latest v2.1.318 Next v2.1.319 Feeds RSS JSON llms.txt llms-full.txt Unofficial

Claude Code v2.1.273 ·

Served model catalog fetch reworked to support API-key and token credentials

Fetching the served model list now works for token/API-key logins, not just full organization accounts, with clearer refusal reasons

Group of 3 You'll notice Improvements
JSON All of v2.1.273
You'll noticeTier: how much it should matter to you
4Useful: my rating, 1 to 5
2Signal: worth watching, 1 to 5
ModelsArea: what it touches
ImprovementsKind: in v2.1.273,
What probably matters to youSection of the release

What

The "served model catalog" (the list of models Claude Code is allowed to offer) used to only be fetched for full organization accounts, hitting an organization-specific endpoint. Now:

  • The fetch branches on the kind of credential in use (organization account, token, or API key), and for token/API-key logins it hits a different, org-less endpoint with an added anthropic-client-platform header.
  • A new refusal reason, lane_unavailable, is tracked separately from the existing not_permitted reason, for cases where the org-less route isn't enabled on the server.
  • The fetch logic now tracks a "lane" and a resolution object, logging route and credential details, and a stale cached catalog entry is discarded when a lane_unavailable reason is hit.
  • The old plain servedCatalogNotPermitted boolean was replaced with a richer servedCatalogRefusal object, and a background tab now explicitly resets its catalog state instead of silently doing nothing.

Why

This lets accounts that log in with a token or API key (rather than a full organization account) get an accurate, up-to-date list of available models, and gives clearer diagnostics when the catalog can't be fetched.

See this entry in the whole of v2.1.273 →

Feedback