MiniMax M3.1 Flash free on Vercel AI Gateway through October 20

MiniMax M3.1 Flash, an open-weight multimodal model built for coding and agentic workloads, is available free on Vercel's AI Gateway until October 20, Vercel announced. The model offers a 1M-token context window, positioning it for long documents and extended agent sessions where conversation history and tool output pile up.

An open-weight model for agents and code

According to Vercel, M3.1 Flash posts gains over earlier MiniMax models on complex agentic tasks, coding, and cybersecurity. On the agent side, Vercel points to multi-step business processes and computer use, where the model carries a task across applications as it works. For coding, MiniMax reports its strongest results in application development, with particular strength in frontend interactions, targeted bug fixes, and component tests.

How the free access works

The free rate comes with one condition: it applies only when MiniMax itself serves the request. Because AI Gateway can route a given model to more than one provider, developers have to sort the provider using the order setting in the gateway's provider options so requests land with MiniMax rather than another backend. Once the promotion ends, requests pinned this way will be billed at MiniMax's standard rate. The window is short — the offer closes the day after Vercel's announcement — so anyone wanting to benchmark the model at zero cost needs to move quickly.

Setting it up

Developers can call the model through the AI SDK, the Chat Completions, Responses, and Anthropic Messages APIs, or a coding agent connected to AI Gateway, using the model name minimax/minimax-m3.1-flash. For Claude Code, Codex, Cursor, and other agents, Vercel says to install the latest Vercel CLI and run npx vercel ai-gateway setup, then select the model inside the agent. A model playground is also available for quick experimentation.

The AI Gateway angle

The launch doubles as a showcase for AI Gateway itself. Vercel describes it as a single API for calling models, tracking usage and spend, and configuring retries, failover, and performance tuning that the company says delivers uptime above what individual providers offer on their own. The gateway also ships with built-in custom reporting, Zero Data Retention support, budgets for API keys, and routing rules. Pricing is passed through at provider rates with no markup and no platform fee on inference, including on Bring Your Own Key requests.

Why it matters

A free, time-boxed trial of an open-weight model with a million-token context window is a low-friction way for developers to test how well current open models handle long-context coding and multi-step agent workflows against closed-model incumbents. For Vercel, the promotion funnels developers onto AI Gateway, and its no-markup pricing claim is a direct pitch at anyone currently paying model providers directly. For MiniMax, placement inside tooling like Claude Code and Cursor through the gateway puts an open-weight option in front of developers at exactly the moment they pick a model. The catch is the deadline: after October 20, evaluation costs revert to standard rates.