AI News (2026/6/12): Dark Side of the Moon Open-Sources New Generation Programming Model Kimi K2.7 Code
Executive Summary:
Dark Side of the Moon announced the open-source release of its new generation programming model, Kimi K2.7 Code. The model shows significant improvements in long-context programming, instruction following, and Agent autonomous execution capabilities, with code benchmark test scores improving by 11%-31.5%, token consumption reduced by 30%, and API pricing remaining the same as K2.6.
Dark Side of the Moon Open-Sources New Generation Programming Model Kimi K2.7 Code
News Details
Dark Side of the Moon announced on Friday, June 12, the open-source release of its new generation programming model, Kimi K2.7 Code. Compared to the previous generation K2.6, Kimi K2.7 Code has achieved significant improvements in several key performance metrics, particularly in long-context programming, instruction following, and Agent autonomous execution capabilities. These enhancements are designed to provide developers with more efficient and accurate programming assistance tools.
Key Points
Long-Context Programming: Kimi K2.7 Code can handle longer code segments, supporting up to 8192 tokens of context. This improvement makes the model more adept at handling complex projects and multi-file code.
Instruction Following: The new model has significantly improved its accuracy in understanding and executing instructions, especially in multi-step instructions and complex tasks. Benchmark tests show that Kimi K2.7 Code's accuracy in code generation tasks has increased by 11%-31.5%.
Agent Autonomous Execution Capability: Kimi K2.7 Code introduces a thinking mode that allows the Agent to perform deeper reasoning and decision-making when executing tasks. This feature enables the Agent to complete some programming tasks autonomously without explicit instructions, thereby enhancing development efficiency.
Reduced Token Consumption: By optimizing the model architecture and algorithms, Kimi K2.7 Code has reduced token consumption by 30%. This not only lowers usage costs but also improves the model's response speed.
API Pricing Remains the Same as K2.6: Despite the significant performance improvements, the API pricing for Kimi K2.7 Code remains unchanged, aligning with the previous generation K2.6. This means developers can enjoy more powerful features without increasing their budget.
AI-ALL In-Depth Commentary
The release of Kimi K2.7 Code marks another Leap forward in the performance and efficiency of programming assistance tools. The enhancements in long-context support and instruction following capabilities Make the model better suited for complex development environments and multi-file projects. Particularly, the introduction of thinking mode empowers the Agent with stronger autonomous execution capabilities, which will significantly reduce repetitive tasks for developers and enhance overall development efficiency.
The reduction in token consumption not only lowers usage costs but also improves the model's response speed, a crucial improvement for developers who frequently call APIs. The API pricing remaining the same as the previous generation further strengthens the model's market competitiveness.
Overall, Kimi K2.7 Code not only provides developers with more powerful tool support but also excels in cost control. This open-source project is expected to further drive the application and development of AI in the programming field, fostering the creation of more innovative projects.
