Fast, capable general-purpose model balancing strong performance with cost efficiency. Features extended thinking and excels at coding, analysis, and complex reasoning.
Strong tool use and function calling
Strong reasoning with extended thinking
| Benchmark | Score | Type | Recorded |
|---|---|---|---|
| SWE-Bench | 85.2 | accuracy | 23d ago |
| SciCode | 54.3 | accuracy | 23d ago |
| MMLU-Pro | 87.5 | accuracy | 23d ago |
| Humanity's Last Exam | 43.2 | accuracy | 23d ago |
| GPQA Diamond | 91.1 | accuracy | 23d ago |
| LCR | 82.0 | accuracy | 23d ago |
| LiveCodeBench | 82.4 | accuracy | 23d ago |
Excellent code generation and debugging
Reliable instruction following
Broad multilingual support
Strong creative writing ability
1M token context window
Accepts image inputs