• D.C.
  • BXL
  • Lagos
  • Riyadh
  • Beijing
  • SG
  • D.C.
  • BXL
  • Lagos
Semafor Logo
  • Riyadh
  • Beijing
  • SG


Flagship newsletter icon
From Semafor Flagship
In your inbox, every weekday
Sign up

Anthropic’s latest AI marks shift toward reasoning models

Feb 25, 2025, 7:19am EST
PostEmailWhatsapp
Anthropic CEO Dario Amodei
Kimberly White/Getty Images for TechCrunch/CC BY 2.0

Anthropic unveiled its latest “hybrid reasoning” artificial intelligence model, which appears to be especially good at solving coding problems.

Claude 3.7 Sonnet users can choose between quick responses like those of ChatGPT, or extended chain-of-thought reasoning in which it shows its working. AIs are moving away from the chatbot model, which is verbally fluent but prone to hallucinations — nonsensical outputs — and towards more autonomous and contemplative systems: Anthropic also released Claude Code, which can write its own programs from users’ written or spoken prompts.

The AI race is “incredibly fast,” The Verge reported, and “Anthropic appears to be in the lead” — and the future seems to be single AIs that can do everything, rather than standalone models for different forms of reasoning.

AD