HISTORICAL RELEASE COMPARISON
Claude 3.7 vs Claude 3.5 Sonnet: reasoning and tools
Claude 3.7 added an explicit choice between ordinary responses and extended thinking. It launched alongside the Claude Code research preview.
These are the releases named below, not current product recommendations.
| What to compare | Claude 3.5 Sonnet | Claude 3.7 Sonnet |
|---|---|---|
| Snapshot | June 2024 Sonnet | February 2025 Sonnet |
| Thinking control | No dedicated extended mode at this launch | Standard or extended thinking |
| Companion release | Artifacts preview | Claude Code research preview |
What this means in practice
A coding agent can inspect files and run tests, while a standalone chat answer may only suggest a patch. Compare the full workflow and its cost. Assigning every improvement to the model alone misses the effect of its tools.
Does this cover the October 2024 update?
No. The 3.5 entry here is the original June release. October 2024 Sonnet must be treated as a separate version when comparing published results.
Read the individual model stories
Sources
Developer reports describe their own results. This site has not independently reproduced them.
Anthropic: Introducing Claude 3.5 Sonnet Google Cloud: June 20 launch availability Anthropic: Claude 3.7 Sonnet and Claude Code