M3.1 Flash

MiniMax M3.1 Flash is the fast, cost-efficient multimodal model from MiniMax and a direct successor to M3, delivering substantial gains across complex agentic tasks, coding, and cybersecurity while holding latency and price low. It is built for long-horizon, high-throughput work — autonomous agent loops, large-scale code and document processing, tool-driven research, and multi-step reasoning pipelines — where throughput and cost per token matter as much as raw peak intelligence. It accepts text, image, and video input and returns text, so interface screenshots and error states, architecture and network diagrams, dashboards, charts, and recorded screen captures can be reasoned over directly alongside source code and long textual context. A 1,048,576-token context window lets whole repositories, large document and log sets, long transcripts, or multi-file projects sit in a single call without a retrieval pipeline, while an output ceiling of up to 524,288 tokens supports long single-pass artifacts — complete reports, extended refactors and code generation, or detailed multi-step plans — without truncation or stitching. Reasoning is always on, with selectable effort (low / medium / high / xhigh / max) so callers can trade latency for depth per request. Native function calling and structured (JSON) output make it well suited to agentic tooling that plans, invokes tools, and returns machine-parseable results, and it can drive coding assistants, browser and computer-use agents, and automated workflows that carry a lot of state across many turns. As the Flash tier it is tuned for breadth and volume rather than one-shot hardest-reasoning tasks — bulk triage, large fan-out agent runs, and interactive assistants served at scale — for which the heavier MiniMax models remain the better fit. It is available over the Chat Completions, OpenAI Responses, and Anthropic Messages APIs, so existing OpenAI- and Anthropic-style clients can target it without code changes. Artificial Analysis Intelligence Index: 42.