On February 24, 2025, Anthropic released Claude 3.7 Sonnet, which it described as “the first hybrid reasoning model on the market.” Rather than splitting reasoning into a separate model line, Claude 3.7 Sonnet could either answer immediately or switch into an “extended thinking” mode where it works through a problem step by step before responding, with the thinking visible to the user. Anthropic noted that “extended thinking mode is available on all surfaces except the free Claude tier,” and kept the same $3 / $15 per-million-token pricing as 3.5, with thinking tokens billed as output.
The release also introduced Claude Code, an agentic coding tool that ran in the developer’s terminal as a limited research preview. Anthropic described it as “an active collaborator that can search and read code, edit files, write and run tests, commit and push code to GitHub, and use command line tools,” delegating real engineering work rather than just suggesting completions.
Claude 3.7 Sonnet mattered as a design statement: instead of forcing users to pick between a fast chat model and a slow reasoning model, it merged both behaviors into one model with a budget dial. That hybrid approach, letting a single model spend more compute only when a task warrants it, became a template other labs followed.