Google: Gemma 3 27B

Gemma 3 27B is a multimodal model from Google that accepts both text and image inputs, supports tool use, and offers a 131K-token context window with up to 16K tokens of output per response. It does not include a built-in reasoning mode, and structured output support is unconfirmed. The combination of vision input and tool support makes it suitable for agentic workflows that involve image understanding alongside function calling. At $0.08 per million input tokens and $0.16 per million output tokens, this is a low-cost option worth considering for budget-sensitive applications. However, its benchmark picture is thin: a blended score of 19.9 across only three benchmarks leaves its real-world capability largely unproven relative to more thoroughly evaluated competitors. Buyers who need confirmed broad performance should treat those numbers cautiously and run their own evaluations before committing.