MiniMax model releases

MiniMax-M3.1-Flash-Preview is MiniMax's fast, lower-cost coding model, launched quietly on 2026-09-27 inside the MiniMax Code tool and via MiniMax's Token Plan subscription. Exposed through both OpenAI-compatible and Anthropic-compatible endpoints (API id MiniMax-M3.1-Flash-Preview). Text, image, and video input, text output; a 1,000,000-token context window (no separately published output cap). It targets everyday development work — bug fixes, small features, and the routine end of a coding queue — rather than the hardest problems, and features always-on reasoning with five tunable effort levels (low, medium, high, xhigh, max) that cannot be fully disabled. Closed weights, with no public model card or parameter count disclosed at launch. MiniMax list pricing is $0.30 / 1M input and $1.20 / 1M output (third-party gateways such as CometAPI resell it around $0.24 / $0.96). A Preview release; no first-party benchmark table had been published as of launch.

Native multimodal MiniMax model with a one-million-token context, sparse attention, and agentic coding/cowork positioning.

Open-weight agentic model from MiniMax focused on real-world software engineering, office tasks, tool use, and self-improving training workflows.

Open Apache-licensed hybrid-attention reasoning model with 456B total / 45.9B active parameters and a native 1M-token context.

Open MiniMax generation with MiniMax-Text-01 and MiniMax-VL-01 long-context models.

MiniMax's commercial long-context abab model generation before the open MiniMax-01 and M series.