SpaceXAI: Grok 4.20

Grok 4.20, listed under the x-ai vendor, accepts text, image, and file inputs and offers a 2,000,000-token context window with up to 1,800,000 completion tokens. It supports tool use and reasoning, which suits agent-style workflows and multi-step tasks. Structured output support is not specified in the available data, and open-weights status is also unlisted, so teams relying on either should verify before committing. At $1.25 per million input tokens and $2.50 per million output tokens, the model sits at a moderate price point, and the large output allowance makes long-document and high-volume generation jobs worth costing out. Its benchmark standing is thin: the blended score of 62.0 comes from a single independent benchmark, EQ-Bench, where it scored 55.8. Shortlist Grok 4.20 if very long context and multimodal file handling matter most, but run your own evaluations, since its broader performance remains largely unproven.