Google: Gemma 3n 4B

Gemma 3n 4B is a text-only model from Google with a 32,768-token context window. It does not support tool use, native reasoning, or structured output, and accepts only text as input. The absence of these features keeps the model straightforward: it handles text-in, text-out tasks without the scaffolding that more complex pipelines require. At $0.06 per million input tokens and $0.12 per million output tokens, it sits at the budget end of the pricing spectrum, which is its clearest argument for shortlisting. However, benchmark coverage is thin, with a blended score of 5.9 drawn from only one independent benchmark, so quality estimates should be treated as provisional. Teams running high-volume, low-complexity text tasks on a tight budget may find the price attractive, but anyone needing tools, multimodal input, or confident quality benchmarks should look at better-covered alternatives first.