Skip to content

All the news that matters, in plain English.

Latest
  1. BBC pauses civil service secondment scheme amid impartiality row
  2. Lavrov presses Gulf security plan in Manama before Tehran mission
  3. Cleveland-Cliffs swings to profit and guides 2027 EBITDA above $4 billion
  4. Lithuania calls Starlink base-station arson attempt 'Russia-linked sabotage', boosts critical-site security
  5. Musk pledges one SpaceX share to every US kid after Anthropic commits 5% of stock to public benefit fund
  6. Barbara Windsor's widower Scott Mitchell confirms split from Tanya Franks
  7. AI chip startup Tenstorrent raises $1.5 billion at $6.2 billion valuation
  8. Twelve migrants drown as boats capsize off Greek islands of Samos and Lesbos

Technology

MiniMax launches M3.1 Flash AI model, claims major gains in speed and cost

Chinese AI startup MiniMax has launched an updated M3.1 Flash model, claiming "generational" improvements, nearly tripled inference speed and 70% lower usage costs.

By Tom Ellery · · 1 min read

Editorial illustration of a computer processor chip surrounded by a glowing neural-network mesh
Illustration — Flip News

Chinese AI startup MiniMax has launched an updated version of its AI model M3.1 Flash, it said in a statement on Monday, claiming a major step forward as competition among Chinese AI labs intensifies.

Reuters reports that the company says the new model delivers “generational” improvements in agentic tasks, programming and cybersecurity, while online inference speed has nearly tripled and usage costs have been reduced by 70%.

In its own launch post, MiniMax says M3.1 Flash reaches “the level of leading models” on complex agentic tasks, computer use, software development and cybersecurity. The company cites 73.7% on the DeepSWE v1.1 coding benchmark, 76.2 on BankerToolBench and 82% on CyberGym, adding that it ranks first among Chinese models on OSWorld V2 and Agents’ Last Exam. These figures are company-published and independently unverified.

Pricing is set at $0.15 per million input tokens and $0.60 per million output tokens, with cached input at $0.015 — its most cost-effective offering yet, the company says. The model is available now through the MiniMax API, M Plan and MiniMax Code; model weights will be released later.

More on this topic: all Technology stories

Get Flip News by email

This opens your email app — we add you manually. No account, no spam, no third parties.