Kimi K2-Instruct-0905 by Moonshot AI is a state-of-the-art Mixture-of-Experts (MoE) language model with 1 trillion total parameters and 32 billion activated parameters. It features an extended 256K token context window and is specifically enhanced for agentic coding intelligence and frontend programming tasks. The model demonstrates significant improvements in performance on public benchmarks for coding agent tasks, making it suitable for complex development workflows.
Loading preview...
Kimi K2-Instruct-0905: Enhanced Agentic Coding Intelligence
Kimi K2-Instruct-0905 is Moonshot AI's latest and most capable Mixture-of-Experts (MoE) language model. It boasts a total of 1 trillion parameters with 32 billion activated parameters, designed for high-performance agentic applications.
Key Capabilities
- Advanced Agentic Coding: Demonstrates significant performance improvements on public benchmarks like SWE-Bench and real-world coding agent tasks, including a 69.2% accuracy on SWE-Bench verified.
- Extended Context Window: Features an impressive 256K token context length, a substantial increase from previous versions, enabling better handling of long-horizon tasks.
- Frontend Programming: Offers advancements in both the aesthetics and practicality of frontend programming, enhancing the developer experience.
- Tool Calling: Possesses strong tool-calling capabilities, allowing it to autonomously decide when and how to invoke external tools based on user requests.
Good For
- Software Development: Ideal for developers and agents requiring robust coding assistance, particularly for complex projects and debugging.
- Long-Context Applications: Suitable for tasks that benefit from processing extensive amounts of information, such as codebases or detailed documentation.
- Automated Workflows: Excellent for integrating into automated systems that leverage tool use for dynamic problem-solving and interaction with external services.
Top 3 parameter combinations used by Featherless users for this model. Click a tab to see each config.