MiniMax M2.5
MiniMax · released Feb 12, 2026 · MiniMaxAI/MiniMax-M2.5
- Type
- Open weightsCustom licence
- Params
- 229B
- Context
- 205K
about 154K words of context · download allowed, licence restricts use
Our take
Written Aug 3, 2026MiniMax M2.5 is a large text-only model with 228.7 billion parameters, released in early 2026. Its standout feature is a coding score that sits well above its general chat rating, though a custom restricted licence and lack of multimodal support limit where it fits.
Pick this when coding quality is the deciding factor and you can work within a restricted licence. Use it for long-context text tasks up to 204,800 tokens, or when you need high throughput and can pay for Mara's 206 tokens per second. Skip it if you need image, video or audio support, if you require a permissive Apache or MIT licence, or if general chat and creative writing quality matter as much as coding.
The case for it
- Coding performance is markedly stronger than its general chat ability: a 54.3-point gap on the Arena leaderboard in coding's favour.
- Ten hosted offers give wide provider choice, with competitive entry-level input pricing available from several hosts.
- A high-throughput option exists for latency-sensitive work, at more than eight times the speed of the slowest tracked provider.
The case against it
- Overall chat and creative writing scores trail its own coding mark by 54.3 and 32.6 points respectively.
- Custom restricted licence — not Apache 2.0 or MIT — so check terms before commercial use, fine-tuning or redistribution.
- Text-to-text only; no image, video or audio input or output.
How good is it?
IntelligencePuzzles, maths, exam questions
Arena Text (overall)84th of 143 · 1389.9
CodingWriting and fixing code on its own
Arena Coding76th of 143 · 1444.2
AgenticPlanning, calling tools, staying on task
MiniMax M2.5 is not on Arena Agent (IPS), which is where the rating would come from, so there is no rating here. It is on SWE-bench Verified, in 2nd of 39 with 75.8.
WritingWe do not rate this
Two boards come close and neither tests writing: Arena Creative Writing asks people which of two replies they prefer, and LiveBench Language tests whether a model understood a passage. So we show where MiniMax M2.5 placed and give it no mark out of five.
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, which is why they get no rating.
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?
- Fits in memory
- weights load entirely on the card
- Spills to system RAM
- some weights offload; much slower
- Too large
- will not load even with offload
- est
- size is calculated; the verdict could change by 10%
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.
All 3 sizes here are calculated, not measured. We hold no measured file for this model, so each size comes from the parameter count and every verdict above inherits that uncertainty.
Check against your own machine → · All 71 devices, with every size →
Or rent it from someone else
Cheapest of 17 live listings. Picked at the widest standard context we hold, within one quantisation slice, so the numbers beside it are a price one host actually charges.
- per 1M tokens
- $0.15 in / $0.90 out
- Context served
- 205K
- Throughput
- Not measured
| Provider | In / out per 1M tokens | Context | Throughput | Trains on prompts | Logs prompts | Zero retention |
|---|---|---|---|---|---|---|
| OpenRouter | $0.15 / $0.90 | 205K | not measured | Unknown | Unknown | Unknown |
| DigitalOcean Gradient | $0.23 / $0.90 | 66K | 74 tok/s | No | No | Confirmed |
| Inceptronfp8 | $0.15 / $0.90 | 197K | 57 tok/s | No | No | Confirmed |
| Venice AI | $0.27 / $0.95 | 198K | 67 tok/s | No | No | Confirmed |
| Mara | $0.24 / $0.96 | 197K | 206 tok/s | No | No | Confirmed |
| StreamLake | $0.27 / $1.08 | 200K | 67 tok/s | No | Yesunknown period | Unknown |
| Parasailfp8 | $0.30 / $1.20 | 197K | 81 tok/s | No | No | Confirmed |
| Novita AIfp8 | $0.30 / $1.20 | 205K | 48 tok/s | No | No | Confirmed |
| AtlasCloudfp8 | $0.29 / $1.20 | 197K | 75 tok/s | No | Yesunknown period | Unknown |
| Chutesfp8 | $0.15 / $1.20 | 197K | 24 tok/s | No | Yesunknown period | Unknown |
| Minimaxfp8 | $0.30 / $1.20 | 205K | 72 tok/s | No | Yesunknown period | Confirmed |
| Friendli | $0.30 / $1.20 | 197K | 65 tok/s | No | Yesunknown period | Unknown |
| SiliconFlowfp8 | $0.30 / $1.20 | 197K | 50 tok/s | No | No | Confirmed |
| CoreWeavefp8 | $0.30 / $1.20 | 197K | 81 tok/s | No | No | Confirmed |
| Phala | $0.20 / $1.38 | 197K | 19 tok/s | No | No | Confirmed |
| Minimaxhighspeed tierfp8 | $0.60 / $2.40 | 205K | 38 tok/s | No | Yesunknown period | Unknown |
| Novita AI | $0.60 / $2.40 | 205K | not measured | Unknown | Unknown | Unknown |
DigitalOcean Gradient
Novita AIAcross the 17 listings we hold: 15 say they do not train on prompts, 0 say they do and 2 do not say. 10 appear in the zero-retention registry we check; the rest are unknown to us rather than confirmed either way.
What each host's API supports
From the parameter list each endpoint publishes. Streaming is omitted: nothing we hold reports it, for any model.
- ✓
- Supported
- ✗
- Not supported
- Not published
- host gave no parameter list
| Provider | Tool calling | JSON output | Strict schema |
|---|---|---|---|
| OpenRouter | ✓ | ✓ | ✓ |
| DigitalOcean Gradient | ✓ | ✓ | ✓ |
| Inceptronfp8 | ✓ | ✓ | ✓ |
| Venice AI | ✓ | ✗ | ✗ |
| Mara | ✓ | ✓ | ✗ |
| StreamLake | ✓ | ✓ | ✓ |
| Parasailfp8 | ✓ | ✓ | ✓ |
| Novita AIfp8 | ✓ | ✓ | ✗ |
| AtlasCloudfp8 | ✓ | ✓ | ✓ |
| Chutesfp8 | ✓ | ✓ | ✓ |
| Minimaxfp8 | ✓ | ✓ | ✗ |
| Friendli | ✓ | ✓ | ✓ |
| SiliconFlowfp8 | ✓ | ✓ | ✓ |
| CoreWeavefp8 | ✓ | ✓ | ✓ |
| Phala | ✓ | ✓ | ✓ |
| Minimaxhighspeed · fp8 | ✓ | ✓ | ✗ |
| Novita AI |
Tool calling: 16 of 17 listings say yes, 1 publishes no parameter list. JSON output: 15 of 17 listings say yes, 1 says no, 1 publishes no parameter list. Strict schema: 11 of 17 listings say yes, 5 say no, 1 publishes no parameter list.
Models people weigh against MiniMax M2.5
When we formed this view
Dates behind this page
Prices last checked 4d ago
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.
- 1 of 17 listings publish no parameter list, so what their API accepts is unknown to us.
- Nothing we hold says whether an endpoint streams, so we do not show it either way.
- 2 of 17 listings do not say whether they train on prompts.
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-M2.5
- Architecture
- Mixture of experts
- Modality record
- text->text
- Catalogue slug
- minimax-minimax-m2-5