Google: Gemma 4 26B A4B (free)

Gemma 4 26B A4B is a multimodal model from Google that accepts image, text, and video inputs. It supports tool use and reasoning, and offers a context window of 262,144 tokens with a maximum of 32,768 output tokens per response. Structured output support is unconfirmed. The model is free to use, with no input or output costs. For teams evaluating cost-sensitive workflows, the zero-dollar price point makes it easy to test at scale without budget risk. That said, there is currently no independent benchmark coverage to gauge where it stands against other models in the same tier. Shortlisting it makes sense if you need multimodal input handling or long-context processing and want to run your own evaluations, but buyers who rely on published benchmarks to make decisions will find the current evidence base insufficient.