Kimi K2.7 Code points to cheaper coding agents
Moonshot released Kimi K2.7 Code as open source this week. Its scores rose from 50.9 to 62.0 on Kimi Code Bench v2, 48.3 to 53.6 on Program Bench, 26.7 to 35.1 on MLS Bench Lite, and 72.8 to 81.1 on MCP Mark Verified. The model stays in the same 1T MoE family, uses 32B active parameters, and supports a 256k context.
The more practical change is a 30% drop in reasoning token use compared with K2.6. That matters for coding agents because they often need to inspect a problem, edit code, run tests, fail, and try again many times. The model still appears behind GPT-5.5 and Opus on coding benchmarks.
But in Moonshot’s MCP Mark Verified table, K2.7 scores 81.1 while Opus 4.8 scores 76.4, which suggests it may already be strong for agent-style coding tasks. A coming high-speed mode claims about 5 to 6 times faster output from the same model, which could make it useful for lowering the time and cost of repeated coding work rather than replacing the best frontier model everywhere.
Key points
- Kimi K2.7 Code was released as open source by Moonshot.
- Moonshot reports a 30% cut in reasoning token use versus K2.6.
- Scores improved across several coding benchmarks, including MCP Mark Verified rising from 72.8 to 81.1.
- Moonshot’s table shows K2.7 ahead of Opus 4.8 on MCP Mark Verified, though vendor numbers need caution.
- A planned high-speed mode claims roughly 5 to 6 times faster output from the same model.
Quick term guide
- open source
- Software whose code is available for people to view and often modify.
- active parameters
- The part of the model that is actually used for a given answer.
- reasoning token
- A small piece of text the model uses while working through an answer, which can add to cost.
- coding agents
- AI programs designed to autonomously perform tasks like writing or fixing code.
- coding agent
- An AI tool that writes or edits code from a person’s instructions.
- agent-style
- A system that tries to complete several steps toward a goal with less manual input.
- frontier model
- The most capable, cutting-edge AI model available at a given time, usually also the most expensive.
- reasoning tokens
- Tokens a model uses for its internal thought process before answering — invisible in the output but included in the bill.