Kimi K2 Instruct vs Qwen3.5 4B
25 SHARED BENCHMARKSAcross 25 shared benchmarks, Kimi K2 Instruct scores higher on 14 and Qwen3.5 4B on 9, with 2 level. The widest gap is OmniScience Non-Hallucination, where Kimi K2 Instruct scores 30.3 against 13.4. Qwen3.5 4B is the cheaper of the two on tracked API pricing ($0.03 against $0.60 per million input tokens).
MOONSHOTVSALIBABA25 SHARED14–9 HEAD-TO-HEAD
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 (direct, direct), otherwise the lowest tracked offer.