MiniMax M2
MiniMax · released Oct 22, 2025 · MiniMaxAI/MiniMax-M2
- Type
- Open weightsCustom licence
- Params
- 229B
- Context
- 205K
10B active per word · about 154K words of context · download allowed, licence restricts use
Our take
Written Sep 17, 2026MiniMax M2 is a text-only model you can download, with a request capacity large enough that long documents need not be split up first. Its custom licence puts conditions on commercial use and redistribution, and the quality evidence is a set of human-preference arena scores plus one coding benchmark run inside a named harness.
Use it for work where you can judge the output yourself and want a downloadable model that takes a long document in one request, or through a host if you would rather not run it yourself. Read the licence before you build on it. Skip it if you need terms that allow commercial use without conditions, or if you need measured quality on a task other than the one the SWE-bench harness covers.
The case for it
- A long report or a stack of documents fits beside the question in one request, though reliable recall across all of it is unverified in our data.
- 61% of real GitHub issues resolved end-to-end on SWE-bench Verified, measured inside the mini-SWE-agent harness, so the figure is for the model in that harness rather than on its own.
- Human preference is measured across several prompt types, with coding the highest of the set and creative writing the lowest; these record which answer people preferred, not whether it was correct.
The case against it
- The custom licence puts conditions on commercial use and redistribution, so it needs reading before you build on it.
- 228.7 billion parameters in total with no active-per-token figure, so memory in use cannot be judged from the size alone.
- Five hosted offers list rates but no measured speed, so price alone cannot pick the host.
How good is it?
An open text model for general chat and code, though it trails most models on everyday questions, drafting and coding.
- getting answers to everyday questionsArena Text (overall) · 128th of 168
- drafts, rewrites and editingArena Creative Writing · 137th of 168
- writing and completing codeArena Coding · 129th of 168
EverydayGeneral questions and everyday reasoning
Arena Text (overall)128th of 168 · 1343
CodingWriting and fixing code on its own
Arena Coding129th of 168 · 1382
AgenticPlanning, calling tools, staying on task
Not yet scored on Arena Agent. It is on SWE-bench Verified, in 22nd of 42 with 61.
WritingDrafting and rewriting prose
Arena Creative Writing137th of 168 · 1285
Arena Creative Writing is the only board that has scored it for this.
These tests check whether a model follows instructions — a precondition for all the work above, but not a measure of how well that work is done.
Every published score for this model8 scoresEvery figure we hold, from 8 boards, with who ran it and a link to the source — including the boards no rating above is built on.
Can you run it yourself?
GeForce RTX 4090 · 24 GB
Too large for this card. The weights do not fit even with part of them offloaded to system memory.
Apple M1 Pro (16-core GPU) · 32 GB
Too large for this card. The weights do not fit even with part of them offloaded to system memory.
Comfortable fit
Apple M3 Ultra (80-core GPU) · 512 GB
Room to spare. 233.8 GB spare means a 10% error in the size would not change the answer.
Memory use by level
Against a 24 GB card.
What is quantisation? →This model on every device we track71 devicesThe Q4 build most people download, on each device: what the weights come to, how much context the memory leaves, and whether it runs. Smallest device that runs it first.
Check against your own machine → · Where to rent it hosted →
Or rent it from someone else
Prices checked 4 hours ago — each listing carries its own date.
The only listing at 205K of context — the other 3 in the table below are not like-for-like. One cheaper row there is outside that comparison: a different quantisation.
- per 1M tokens
- $0.30 in / $1.20 out
- Context served
- 205K
- Throughput
- Not measured
| Provider | In / out per 1M tokens | Context | Throughput | Trains on prompts | Logs prompts | Zero retention |
|---|---|---|---|---|---|---|
| Minimaxfp8Through OpenRouter | $0.26 / $1.02checked 4 hours ago | 205K131K max reply | 66 tok/s | No | Yesunknown period | Confirmed |
| OpenRouterOpenRouter's own listing | $0.30 / $1.20checked 4 hours ago | 205K | not measured | Unknown | Unknown | Unknown |
| Novita AIfp8Direct and through OpenRouter | $0.30 / $1.20checked 4 hours ago | 205K131K max reply through OpenRouter | 67 tok/sthrough OpenRouter | DirectUnknownThrough OpenRouterNo | DirectUnknownThrough OpenRouterNo | DirectUnknownThrough OpenRouterConfirmed |
| Google Vertex AIThrough OpenRouter | $0.30 / $1.20checked 4 hours ago | 197K177K max reply | 68 tok/s | No | No | Confirmed |
Novita AI
Google Vertex AIAcross the 4 listings we hold: 3 say they do not train on prompts (1 of them only through OpenRouter), 0 say they do and 1 does not say. 3 appear in the zero-retention registry we check (1 of them only through OpenRouter); the rest are unknown to us.
What each host's API supports
From the parameter list each endpoint publishes. Streaming is omitted: nothing we hold reports it, for any model.
| Provider | Tool calling | JSON output | Strict schema |
|---|---|---|---|
| Minimaxfp8Through OpenRouter | ✓ | ✗ | ✗ |
| OpenRouterOpenRouter's own listing | ✓ | ✓ | ✓ |
| Novita AIfp8Direct and through OpenRouter | ✓ | ✓ | ✗ |
| Google Vertex AIThrough OpenRouter | ✓ | ✓ | ✓ |
Tool calling: 4 of 4 listings say yes. JSON output: 3 of 4 listings say yes, 1 says no. Strict schema: 2 of 4 listings say yes, 2 say no.
Models people weigh against MiniMax M2
When we formed this view
Recent changes
What moved
first indexed by our pipelineEach date is the day we first saw the change, or the day the maker announced it.
What we do not know about this model yet
- We hold no measured file for it, so all 3 sizes on this page are calculated from the parameter count.
- Nothing we hold says whether an endpoint streams, so we do not show it either way.
- 1 of 4 listings does not say whether it trains on prompts, and 1 answers only through OpenRouter, not for its own listing.
- We hold no batch or off-peak rate for any of its listings.
- We hold a decode speed for it, but no prompt-processing (prefill) figure, so how long the input side of a job takes is unknown to us.
Licence and identifiers
What the licence allowsCustom licence, what it allows commercially, and the identifiers you need to pull this model — its Hugging Face repo, our slug and a machine-readable card.
Licence
Custom licence
This model ships custom license terms that don't map to a known template. We haven't parsed them, so commercial use, redistribution and derivatives are unverified — review the original terms before shipping.
Identifiers
- Hugging Face
- MiniMaxAI/MiniMax-M2
- Architecture
- Mixture of experts
- Takes in, gives back
- Text in, text out
- Catalogue slug
- minimax-minimax-m2