Google: Gemma 4 26B A4B

Google's Gemma 4 26B A4B accepts text, image, and video input and returns text. It offers a 262,144-token context window with up to 235,929 completion tokens, so it can take long documents or extended conversations and produce lengthy outputs in a single call. It supports tool calling and a reasoning mode, which suits agent-style workflows. Structured output support is not listed, so verify it before relying on schema-constrained responses. The main reason to shortlist this model is cost: $0.0675 per million input tokens and $0.225 per million output tokens puts it at the low end for a multimodal model with tools and reasoning. Its benchmark evidence is thin. It has a blended score of 49.4 drawn from only two benchmarks: 27.5 on the AI Index and 70.0 on EQ-Bench. That gap suggests stronger conversational and emotional-intelligence performance than general capability, though two data points are limited. Teams processing high volumes of mixed media on a budget should test it against their own tasks before committing.