MiniMax Launches M3.1 Flash — Claims 3x Faster Inference and 70% Lower Cost

A new AI model just dropped, and MiniMax is making some big promises about it. M3.1 Flash is the company's latest "Flash" model — a category that usually means the faster, lighter cousin of a full-sized AI. Think of it like the difference between a phone's battery-saver mode and its high-performance mode.

What MiniMax says it does better

According to the announcement, M3.1 Flash delivers a "generational leap" in four specific areas. Here's each one in plain English.

• Agentic tasks. An "agent" is an AI that can take actions on its own — like opening a browser, clicking buttons, or filling in a form — instead of just answering questions. MiniMax says M3.1 Flash handles this multi-step work much better.
• Coding. AI that writes or fixes software. According to MiniMax, this model is a noticeable step up for code-related tasks.
• Computer Use. A specific AI skill where the model looks at a screen and controls a computer the way a person would — moving the cursor, opening apps, typing. MiniMax claims stronger performance here too.
• Cybersecurity. Tasks related to finding or fixing security problems in software.

The speed and cost story

The two numbers MiniMax highlights most are about efficiency, not raw intelligence.

• Online inference — which just means "the moment the AI is actively processing your request" — is "nearly three times faster" than before, the company says.
• Usage costs are cut by 70%, also per MiniMax's own claim.

These are the company's own figures, not independent benchmarks. M3.1 Flash is rolling out today, so it will take time before outside reviewers can verify how those numbers hold up in the real world.