MiniMax: MiniMax M2.1

MiniMax M2.1 is a text-in, text-out model from MiniMax with a 204,800-token context window and a large maximum output of roughly 196,000 tokens. It supports tool use and reasoning, making it capable of multi-step and agentic workflows. Structured output support is unconfirmed based on available data. At $0.29 per million input tokens and $0.95 per million output tokens, it sits at a low price point relative to many reasoning-capable models. Its blended benchmark score of 63.9 draws from only three benchmarks, so the picture is limited; the agentic score of 78.2 is its clearest signal of strength, while a coding score of 54.1 suggests it is not the obvious pick for code-heavy tasks. Teams prioritizing long-context agentic pipelines on a tight budget should shortlist it, but those needing broader validation or strong coding performance should weigh the thin benchmark coverage carefully.