Qwen2.5 Instruct 72B vs Qwen3.5 4B
19 SHARED BENCHMARKSAcross 19 shared benchmarks, Qwen2.5 Instruct 72B scores higher on 5 and Qwen3.5 4B on 13, with 1 level. The widest gap is Terminal-Bench Hard, where Qwen3.5 4B scores 18.2 against 4.5. Qwen3.5 4B is the cheaper of the two on tracked API pricing ($0.03 against $0.47 per million input tokens).
ALIBABAVSALIBABA19 SHARED5–13 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.