Across 29 shared benchmarks, Gemma 4 26B A4B scores higher on 13 and Qwen3 VL 235B A22B Reasoning on 15, with 1 level. The widest gap is Terminal-Bench Hard, where Gemma 4 26B A4B scores 25 against 11.4. Gemma 4 26B A4B is the cheaper of the two on tracked API pricing ($0.13 against $0.40 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 (direct, alibaba-official), otherwise the lowest tracked offer.