Across 25 shared benchmarks, DeepSeek-V4-Flash scores higher on 24 and Kimi K2 Instruct on 1. The widest gap is browsecomp, where DeepSeek-V4-Flash scores 73.2 against 14.1. DeepSeek-V4-Flash is the cheaper of the two on tracked API pricing ($0.44 against $0.60 per million input tokens).
Best tracked score per model per benchmark (default configuration; source-attributed). ↓ marks lower-is-better metrics. Open either model for its full surface, provenance and pricing. Quoted rates are the price-setter row we currently track for each model — its direct or vendor-official listing where one exists (deepseek-official, direct), otherwise the lowest tracked offer.