AI Briefing
KO

Claude Adaptive Thinking Becomes Default

·2026.05.15 02:03

Key point

Anthropic has deprecated manual Extended Thinking for Claude Opus 4.7.

Details

The extended thinking policy for the Claude API has changed by model.

  • In Claude Opus 4.7, thinking: {type: "enabled", budget_tokens: N} is no longer supported and returns a 400 error.
  • Instead, you must use thinking: {type: "adaptive"} along with the effort parameter.
  • In Claude Opus 4.6 and Claude Sonnet 4.6, adaptive thinking is recommended; the existing manual extended thinking will continue to work for now but is deprecated.
  • budget_tokens-based configuration is also deprecated for both models and will be removed in a future model release.
reasoningapianthropicclaudemodel-update
Read originalShare on Threads

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.

Related

Hacker NewsUS Appeals Court Upholds Pentagon Blacklisting of Anthropic as Supply Chain RiskRedditAnthropic Launches Claude Docs for Native In-Conversation Document CollaborationAnthropic NewsAnthropic Discovers 'ART', a Novel CRISPR-like Enzyme System, Using ClaudeRedditUiPath Reveals Performance Gap Between Claude Models Based on Agent Task LengthAnthropic NewsCountering AI Misuse: September 2026TLDR AIAnalysis of the 'Inference Gap' Between Claude's Reduced Reasoning Tokens and Benchmarks