Claude Code bug leaves Opus 4.8 sessions running without thinking on xhigh WebSearch
sp4mmami · ghdev · 2026-07-21
Bug report: Claude Code fails to enable adaptive thinking on Opus 4.8
A GitHub issue against anthropics/claude-code says the desktop app on Opus 4.8 does not translate alwaysThinkingEnabled: true into thinking: {type: "adaptive"} on outgoing requests.
What breaks
- On Opus 4.8, adaptive thinking is not enabled by default.
- Because the client omits the thinking field, sessions can run silently without extended thinking even when the user expects it.
- At xhigh/max effort, a server-side tool call such as WebSearch fails with:
- outputconfig.effort 'xhigh' is not supported when thinking is disabled on this model
Repro and scope
- Environment: Claude Code Desktop for Windows, Max 5x plan, direct Anthropic connection
- Settings: alwaysThinkingEnabled: true, model: "opus", effortLevel: "medium"
- Switching the same session to Fable 5 works, because adaptive thinking is always on at the API level there
The reporter includes a verified matrix and argues the bug is isolated to client-side request construction for Opus 4.8.
More from coding & agent
- 'My intelligence stack': developer splits AI coding work across Opus 5.5, Fable and Astra with human review gates — sh_reya · 2026-10-27
- Baseten webinar: evaluate open models for coding agents by benchmarking on your own merged PRs — baseten · 2026-10-27
- Thorsten Ball on working with agents: run more evals, automate everything, use the best models — omarsar0 · 2026-10-27
- DeepMind paper: LLMs can't sample randomly on their own — pass random numbers from code instead — rohanpaul_ai · 2026-10-27
- BELL: Open-Source Stack Combines Browserbase, Eve, iMessage and Stripe Link for Personal Agents — jeff_weinstein · 2026-10-27
- Legora runs 2,000 parallel agent evals on Modal Sandboxes for legal AI — blelbach · 2026-10-27