MiniMax M3.1 Flash Is Free on AI Gateway Until Oct 22

Vercel says MiniMax M3.1 Flash is available through AI Gateway for free until Oct 22. However, this offer has an important condition for people who want to try it. The request must be handled by MiniMax, rather than another provider. A provider is the service that runs the model and returns its answer.

MiniMax M3.1 Flash is an open-weight model for coding and agent tasks. Open weight means the model's trained weights are available, but the article does not describe its license. The model has a 1M-token context window for long documents and sessions. A context window is the amount of material a model can work with at once. Agent tasks can include several steps across different applications.

MiniMax reports improvements over its earlier models in complex agent work, coding, and cybersecurity. Its reported coding strengths include app development, frontend interactions, focused bug fixes, and component tests. These are the provider's claims about the model. The announcement does not present an independent comparison of the results. Readers should keep that difference in mind when considering the reported improvements.

Users can connect through AI SDK or supported APIs and coding agents. The model name is minimax/minimax-m3.1-flash. To receive the free rate, users need to set the provider order in gateway options. After the offer ends, requests pinned to MiniMax use its standard rate. The article does not give an exact ending time or timezone.

AI Gateway lets users call models through one API and track usage and cost. It also includes controls for retries, routing, and switching providers when needed. Users can set budgets for API keys and use custom reports. Vercel says model prices have no added markup or platform fee for inference. The free offer and these gateway features are separate parts of the announcement.