Google: Gemma 4 26B A4B (free)

Gemma 4 26B A4B (free) is a Google model that accepts text, image, and video input. It offers a 262,144-token context window with up to 32,768 output tokens per response, enough to handle long documents or extended transcripts in a single request. It supports tool calling and reasoning. Structured output support is not listed, so teams that need strict schema adherence should test before committing. The main draw is cost; input and output are both priced at zero, which makes it a reasonable shortlist candidate for prototyping, high-volume experiments, or multimodal workloads where budget is the binding constraint. Its benchmark record is thin, with a blended score of 49.4 from only two independent benchmarks: 27.5 on the AI Index and 70.0 on EQ-Bench. That gap points to stronger conversational results than general capability, but with coverage this limited, treat its standing as unproven and validate it on your own tasks.