Claude Code Brings Fast Mode to Vertex and Falls Back to Standard Speed on the Same Model
Claude CodeClaude Code 2.1.322 adds fast mode on Vertex in the global, us and eu regions, and falls back to standard speed if Vertex refuses it. A fix in 2.1.321 changes what happens when fast serving is rate-limited, overloaded or failing: Claude Code now tries standard speed on the same model first, instead of losing replies, switching models or retrying for minutes.
Key Takeaways
- Fast mode now works on Vertex in the global, us and eu regions.
- A refused fast request falls back to standard speed, so the call completes.
- 2.1.321 changed the failure path: standard speed on the same model comes first.
- Replies are no longer lost, and models no longer switch silently, when fast serving is overloaded.
- Long retry stalls are replaced by an immediate fallback.
- Labels in
/fast,/configand/modelname the active model, so users can see which model fast mode applies to.
Fast mode on Vertex
Claude Code 2.1.322 extends fast mode to Google Cloud's Vertex platform. It works in the global, us and eu regions. If Vertex refuses a fast mode request, Claude Code falls back to standard speed, so the request still completes.
A gentler fallback when fast serving fails
In 2.1.321, Anthropic changed what happens when fast serving is rate-limited, overloaded or failing. Before the fix, fast mode could lose replies, switch to a different model, or retry for minutes. Claude Code now tries standard speed on the same model first. The model stays the one the user picked, and the reply arrives at a normal pace, instead of a long stall.
Clearer labels
The fast mode labels in /fast, /config and /model now name the current model when it supports fast mode, and the "(Opus only)" wording was removed from the VS Code row. Fast mode is also available with Claude Fable 5.2, as covered in a separate entry.