Kimi K2-Instruct-0905
MoonshotAIOverview
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.
Kimi K2-Instruct-0905 was released on September 5, 2025.
Performance
Timeline
Other Details
Related Models
Compare Kimi K2-Instruct-0905 to other models by quality (GPQA score) vs cost. Higher scores and lower costs represent better value.
Performance visualization loading...
Gathering benchmark data from similar models
Benchmarks
Kimi K2-Instruct-0905 Performance Across Datasets
Scores sourced from the model's scorecard, paper, or official blog posts
llm-stats.com - Sun Dec 14 2025Pricing
Pricing, performance, and capabilities for Kimi K2-Instruct-0905 across different providers:
Example Outputs
Recent Posts
Recent Reviews
API Access
API Access Coming Soon
API access for Kimi K2-Instruct-0905 will be available soon through our gateway.
FAQ
Common questions about Kimi K2-Instruct-0905