Running Your Business
Cheaper AI Models Keep Coming: How to Choose One

On October 19, MiniMax released M3.1 Flash, a fast, lower-cost AI model. It first appeared as a preview in MiniMax Code and is now generally available.
What it is for
MiniMax built it for coding, for using a computer, and for long jobs done by AI helpers. It says jobs cost about 70% less than with its last model, and that it runs at over 200 tokens (small chunks of text) per second.
Price
For every million tokens:
- $0.15 to read
- $0.015 to read again from memory (cached)
- $0.60 to write
You can use it through the MiniMax API, its M Plan subscription, or MiniMax Code. Downloadable weights are not out yet. They will come later with a larger Pro model.
What to keep in mind
The speed and cost claims come from MiniMax. Its test scores also trail some rivals on some tests.
A simple checklist for choosing an AI model
- Price. What does it cost for your usual amount of work?
- Speed. Is it fast enough for your customers?
- Quality. Does it do well on a real task from your business?
- Your data. Where is it stored and who can read it?
If you pay for AI tools, ask whether a cheaper model could do the same job for you.
Sources
Want a website your customers can trust, kept safe and up to date? Get a free quote.
Questions about this topic
Is M3.1 Flash really 70% cheaper?
That is MiniMax's own claim for jobs run by AI helpers. Check it against your own work.
Can I download it?
Not yet. MiniMax says the downloadable version comes later.


