Kimi K2-Instruct-0905
Moonshot AI · open weight · kimi-k2-instruct-0905
Kimi K2-Instruct-0905 is the latest, most capable version of Kimi K2, achieving state-of-the-art performance in frontier knowledge, math, and coding among non-thinking models. This Mixture-of-Experts model features 32 billion activated parameters and 1 trillion total parameters, meticulously optimized for agentic tasks. Key features include enhanced agentic coding intelligence, extended context length to 256K tokens, and a hybrid architecture trained with MuonClip optimizer on 15.5T tokens. The model achieves 65.8% on SWE-bench Verified (single attempt), 47.3% on SWE-bench Multilingual, and excels at tool use with 70.6% on Tau2-retail. It is a reflex-grade model without long thinking, designed to act and execute complex tasks seamlessly.
Benchmark scores
| Benchmark | Score |
|---|---|
| MATH-500 | 0.97 |
| MMLU-Redux | 0.93 |
| IFEval | 0.90 |
| MMLU | 0.90 |
| MMLU-Pro | 0.81 |
| GPQA | 0.75 |
| AIME 2024 | 0.70 |
| SWE-Bench Verified | 0.66 |
| LiveCodeBench | 0.54 |
| AIME 2025 | 0.49 |
| Humanity's Last Exam | 0.05 |
Pricing
- No provider pricing.
AA metrics
No Artificial Analysis link yet.
Arena Elo
- No Arena snapshot linked.
