Friday, February 27, 2026Model UpdateHigh Impact
ClaudeVersion Opus 4.6

Anthropic Releases Claude Opus 4.6 with Extended Thinking

Anthropic launched Claude Opus 4.6, featuring an upgraded extended thinking mode that allows the model to reason through complex problems for up to 128K tokens before responding. The model leads on SWE-bench Verified at 62.4%.

Key Highlights

1

128K token extended thinking capability

2

62.4% on SWE-bench Verified (new state of the art)

3

96.3% on HumanEval code generation

4

Improved safety scores across all benchmarks

← Back to changelog