MiniMax-M3.1-Flash-Preview

MiniMax-M3.1-Flash-Preview debuted on September 27, 2026, initially inside MiniMax Code, positioned for everyday development work spanning bug fixes through full feature delivery. A first-hand test wired it into OpenCode via a custom provider entry and confirmed it answers on the public API, softening the walled-garden impression from the launch note. The author collected a day of speed and pass-rate numbers against the model directly. Beyond benchmarks, the testing surfaced one consistent failure mode so repeatable it could be used as a timing signal. MiniMax also layered promotional sweeteners on launch: Token Plan quotas reset at launch and daily check-in credits doubled from September 28 through October 7, applicable across the catalog including H3 video generation. No published comparison to MiniMax M3 was supplied at launch.

BenchLM confirms the September 27, 2026 release of MiniMax M3.1 Flash Preview as a tracked proprietary model. The preview is available inside MiniMax Code, but no public benchmark score has been published yet, and no API model ID, context window, maximum output, knowledge cutoff, modalities, or parameter count has been disclosed by the provider. Open-weight availability also remains unannounced. The tracker notes that tool calling, structured outputs, batch support, prompt caching, and rate limits are not yet documented for the preview tier. Lifecycle fields such as cloud regions are similarly untracked. A BenchLM aggregate score will only appear once compatible public benchmark evaluations land for this model.

MiniMax rolled out M3.1-Flash-Preview on September 27, 2026, integrating the model directly into the MiniMax Code developer platform rather than releasing a standalone API. The model targets everyday engineering workflows including bug fixes, refactoring, and end-to-end feature implementation. Because it operates natively on project files, developers avoid context-switching between an external chatbot and their editor. A key product feature is a five-tier reasoning effort selector labeled Low, Medium, High, XHigh, and Max. Lower tiers handle lightweight syntax edits with near-instant responses, while higher tiers support architectural reviews and multi-step refactoring that demand deeper inference. This tiered control lets developers balance latency against analytical depth depending on task complexity.

MiniMax officially launched its M3.1-Flash-Preview text model on September 27, 2026, making it available inside the MiniMax Code platform for day-to-day development work. The model is positioned for routine bug fixes and full feature implementation, with the ability to interpret complex requirements, handle edge cases, and automatically run regression testing. This connects problem identification, code implementation, testing, and delivery into a single development loop for developers. Alongside the launch, MiniMax announced platform-level benefits including a Token Plan quota reset card and a double-points check-in promotion running September 28 through October 7. These incentives are designed to lower developer costs during the preview period. The preview label signals an early, lightweight coding-focused release rather than a final specification for the broader M3.1 family.