MiniMax M2.1 (MiniMax)

MiniMax M2.1 is positioned as a large-scale language model focused on software engineering tasks, and a third-party benchmark tracker lists it with 230 billion total parameters and an advertised 1 million-token context window. The same listing scores the model at 0.261 on OctoCodingBench, a text benchmark designed to evaluate multi-language programming capabilities, where it is shown as the sole tracked entry. Because the OctoCodingBench page is a third-party leaderboard rather than an official MiniMax release, the 230B figure and the 1M context figure should be read as reported rather than as confirmed creator disclosures.

In practical terms, the leaderboard profile suggests a model oriented toward multi-language code generation and refactoring with a very large working context for long repository or multi-file tasks. The combination of a heavy parameter count and an extended context window is consistent with workflows where large codebases, build logs, or extended documentation need to be processed in a single session. Given that the official MiniMax API documentation page visible in the supplied evidence enumerates newer and legacy siblings but does not name M2.1 directly, the model appears to sit in the broader MiniMax coding family as a high-capacity option for developers prioritizing coding tasks at scale.