Kimi K2.7 Code
Moonshot AI's coding-focused agentic model built on Kimi K2.6, delivering stronger long-horizon software engineering performance with roughly 30% lower thinking-token usage.
Moonshot
DeepInfraKimi K2.7 Code
**Kimi K2.7 Code** is a coding-focused agentic model built upon Kimi K2.6. It delivers substantial improvements on real-world long-horizon coding tasks, strengthening end-to-end task completion across complex software engineering workflows while improving token efficiency. ### Key Highlights - **Agentic coding specialization**: Optimized for autonomous, multi-step software engineering tasks that require sustained reasoning and tool use across long horizons. - **Improved token efficiency**: Reduces thinking-token usage by approximately **30% compared with Kimi K2.6**, delivering faster and more cost-effective completions without sacrificing quality. - **End-to-end task completion**: Strengthened performance on complete software engineering workflows — from planning and implementation to debugging and refinement. - **Built on Kimi K2.6**: Inherits the strong general capabilities of its predecessor while adding targeted improvements for coding-heavy use cases. Kimi K2.7 Code is best suited for developers, coding agents, and engineering platforms that need a reliable, efficient model for autonomous software development, code generation, refactoring, and complex debugging tasks.
Ready to build with Kimi K2.7 Code?
Get Started FreeParameters & options
An alternative to sampling with temperature, called nucleus sampling, where the model considers the results of the tokens with top_p probability mass. So 0.1 means only the tokens comprising the top 10% probability mass are considered.
Float that represents the minimum probability for a token to be considered, relative to the probability of the most likely token. Set to 0 to disable this.
Sample from the best k (number of) tokens. 0 means off.
Positive values penalize new tokens based on whether they appear in the text so far, increasing the model's likelihood to talk about new topics.
Positive values penalize new tokens based on how many times they appear in the text so far, increasing the model's likelihood to talk about new topics.
Alternative penalty for repetition, but multiplicative instead of additive (> 1 penalize, < 1 encourage).
A sequence where the API will stop generating further tokens.
Explore similar models
MoonshotKimi K3
MoonshotKimi K2.6
Kimi K2.6 is an open-source, native multimodal agentic model that advances practical capabilities in long-horizon coding, coding-driven design, proactive autonomous execution, and swarm-based task orchestration.
MoonshotKimi K2.5
Moonshot AI's most powerful open-source model — a 1T-parameter native multimodal agent capable of visual understanding, advanced coding, and coordinated multi-agent task execution.
GoogleGemini 3.1 Pro
Google's frontier reasoning model delivering enhanced software engineering, agentic reliability, and multimodal intelligence across a 1M-token context window.
OpenAIGPT‑5.2 Pro
The most capable model series yet for professional knowledge work.
AnthropicClaude 5 Opus
Anthropic's most intelligent model, setting new standards in agentic coding, complex reasoning, and professional knowledge work.
Start building with Kimi K2.7 Code
No API keys required. Create AI-powered workflows with Kimi K2.7 Code in minutes — free.