# Claude always-thinking models no longer return 400s or empty answers

> The Anthropic driver rejects prefill early, maps an explicit none effort to low, and keeps requests append-only on Opus 5.5 and other always-thinking models.

Source: <https://everruns.com/docs/changelog/2026-09-23-claude-always-thinking-fixes/>

Date: 2026-09-23
Products: Platform, Cloud, Framework
Category: Fix
Available: In 0.31.0
Pull request: everruns/everruns#3782
Pull request: everruns/everruns#3783

Requests to Claude models that always think, such as Opus 5.5, now behave predictably:

* A request that ends on an assistant turn (prefill) fails fast with a configuration error that names the fix, instead of a provider 400 and a retry cycle.
* An explicit `none` reasoning effort maps to `low` on models where thinking cannot be turned off. Before, the driver sent nothing and the API ran at its own default.
* Each request in a session replays the previous one and only appends to it, so preserved thinking stays valid across turns.

See [Anthropic](https://everruns.com/docs/providers/anthropic/). PRs [#3782](https://github.com/everruns/everruns/pull/3782), [#3783](https://github.com/everruns/everruns/pull/3783).