Here comes Kimi K2.7 Code: Better Coding with more efficiency

Announcement
1

:tada: :tada: Here comes Kimi K2.7 Code: Better Coding with more efficiency

Benchmarks (SOTA with tools)

  • Kimi Code Bench V2: 62.0 (v.s. K2.6 50.9)

  • Program Bench: 53.6 (v.s. K2.6 48.3)

  • MLS Bench Lite: 35.1 (v.s. K2.6 26.7)

  • Kimi Claw 24/7 Bench: 46.9 (v.s. K2.6 42.9)

  • MCP Atlas: 76.0 (v.s. K2.6 69.4)

  • MCP Mark Verified: 81.1 (v.s. K2.6 72.8)

11
111280×720 125 KB

What’s New

  • Long-horizon coding — Higher success rates on 4,000+ tool call sequences and 12+ hours of continuous execution. Generalization across languages (Rust, Go, Python) and tasks (frontend, DevOps, perf optimization).

  • Optimized reasoning efficiency — Reasoning models tend to overthink, spending thousands of tokens deliberating on problems that don’t need it. Kimi K2.7 Code significantly reduces this tendency: it cuts thinking-token usage by approximately 30% on average compared with K2.6.

  • Multimodal tool use — Native video ingestion with temporal analysis and clip extraction via custom tools. Images (PNG, JPEG, WebP, GIF) and video (MP4, MOV, AVI, etc.) supported.

  • K2.7 Code HighSpeed — Same model capabilities at ~180 tokens/s, peaking at 260 tokens/s in short contexts.

  • Fixed parameters — Temperature (1.0), top_p (0.95), n (1), presence/frequency penalties (0.0) locked for consistent, reproducible outputs.

12
121080×708 63.1 KB

K2.7 Code and HighSpeed is now live on kimi.com in chat mode and agent mode.

For production-grade coding, pair K2.7 with Kimi Code: kimi.com/code

Links

1 Like