Kimi K3 is now available in Devin

Cognition1 min read

Kimi K3 is now live in Devin Desktop and Devin CLI.

On FrontierCode 1.1, our proprietary benchmark that grades models on real-world engineering tasks based on quality and mergeability, Kimi K3 is approaching frontier-level performance, surpassing GPT-5.5, and falls only behind Opus, Fable, and GPT-5.6 Sol. It is the only open source model that achieves this level of performance.

FrontierCode Leaderboard Benchmarks for how well models meet the standards of high-quality production codebases View Now

FrontierCode 1.1 ExtendedScore

020406080100Score (%)40.0Claude Sonnet 4.654.6SWE-1.755.8GPT-5.6 Terra56.2Claude Sonnet 556.7GPT-5.558.2Kimi K359.6Claude Opus 4.860.6GPT-5.6 Sol63.6Claude Opus 564.9Claude Fable 5

Score is a weighted aggregate of rubric items. Solutions that don’t pass blocking criteria receive 0.

FrontierCode 1.1 ExtendedScorevsCost ($)

$0.00$2.00$4.00$6.00$8.00$10$12$14$16avg cost (USD) per rollout35404550556065score (%)Fable 5Opus 5GPT-5.6 SolOpus 4.8Kimi K3Sonnet 5GPT-5.5GPT-5.6 TerraSWE-1.7

Score: a weighted aggregate of the rubric items. Solutions that don’t pass blocking criteria receive 0.

Cost ($): the mean USD spend per rollout. Each dot is a reasoning effort.

In our testing, we see that Kimi K3 specifically excels at debugging. It discovers ground truth by running code, rather than assuming or pattern matching. It often opts to reproduce bugs on its own before editing any files. It also is capable at operating its environment, spinning up required services and cleaning up after itself. However, it falls short in spec adherence. It deviates from stated specs sometimes, an area at which GPT-5.6 Sol, Opus 5 and Fable 5 excel.

Try Devin Cloud · Download Devin Desktop · Install Devin CLI