MiMo-V2.5
Xiaomi · released Apr 27, 2026 · XiaomiMiMo/MiMo-V2.5
- Type
- Open weightsMIT License
- Params
- 311B
- Context
- 1.1M
about 788K words of context
Our take
Written Aug 3, 2026MiMo-V2.5 is a large downloadable model from Xiaomi that accepts text, images, audio and video, and can handle over one million tokens in a single request. Its permissive MIT licence and strong measured coding score make it a notable option for long-document and coding-heavy workloads.
Choose this for long-context document analysis at over one million tokens, or for coding workloads where its measured coding score matters. Good for multimodal pipelines and products needing a permissive commercial licence. Skip it if creative writing quality is central, if you need consistent throughput guarantees, or if you need to know per-token compute cost for self-hosting.
The case for it
- Extremely long request limit for a downloadable model: 1,050,000 tokens.
- Permissive MIT licence allows commercial use, modification and redistribution without copyleft.
- Strong measured coding performance relative to its other skills: 57.5 points above its overall text score.
- Wide provider availability with a roughly fourfold spread on input rates, so shopping around pays off.
The case against it
- Creative writing is a clear relative weakness, sitting 39.6 points below its overall text score and 97.1 points below its coding score.
- No disclosed active-parameter count, so true per-token compute cost for self-hosting is unknown.
- Throughput inconsistent and unmeasured on three of ten tracked offers; hard-prompt strength does not lift overall text rating proportionally.
How good is it?
IntelligencePuzzles, maths, exam questions
Arena Text (overall)50th of 143 · 1433.7
CodingWriting and fixing code on its own
Arena Coding43rd of 143 · 1491.2
AgenticPlanning, calling tools, staying on task
Nobody we watch has scored MiMo-V2.5 for this. We would take the rating from Arena Agent (IPS).
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 MiMo-V2.5 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 model7 scoresEvery figure we hold, from 7 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. 180.7 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 11 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
- $0.14 in / $0.28 out
- Context served
- 1.1M
- Throughput
- Not measured
| Provider | In / out per 1M tokens | Context | Throughput | Trains on prompts | Logs prompts | Zero retention |
|---|---|---|---|---|---|---|
| GMICloudfp8 | $0.11 / $0.22 | 1.1M | 52 tok/s | No | Yesunknown period | Unknown |
| Xiaomifp8 | $0.14 / $0.28 | 1M | 43 tok/s | No | Yes30 days | Unknown |
| Venice AIfp8 | $0.14 / $0.28 | 1M | 46 tok/s | No | No | Confirmed |
| DigitalOcean Gradient | $0.10 / $0.28 | 262K | not measured | No | No | Unknown |
| Parasailfp8 | $0.14 / $0.28 | 1M | 38 tok/s | No | No | Confirmed |
| OpenRouter | $0.14 / $0.28 | 1.1M | not measured | Unknown | Unknown | Unknown |
| Io Netfp8 | $0.21 / $0.32 | 262K | 51 tok/s | No | No | Unknown |
| Novita AI | $0.17 / $0.34 | 1M | not measured | Unknown | Unknown | Unknown |
| Novita AIfp8 | $0.17 / $0.34 | 1M | 46 tok/s | No | No | Confirmed |
| DeepInfrabf16 | $0.40 / $2.00 | 262K | 49 tok/s | No | No | Confirmed |
| DeepInfra | $0.40 / $2.00 | 262K | not measured | Unknown | Unknown | Unknown |
DigitalOcean Gradient
Novita AI
DeepInfraAcross the 11 listings we hold: 8 say they do not train on prompts, 0 say they do and 3 do not say. 4 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 |
|---|---|---|---|
| GMICloudfp8 | ✓ | ✗ | ✗ |
| Xiaomifp8 | ✓ | ✓ | ✗ |
| Venice AIfp8 | ✓ | ✓ | ✓ |
| DigitalOcean Gradient | |||
| Parasailfp8 | ✓ | ✓ | ✓ |
| OpenRouter | ✓ | ✓ | ✓ |
| Io Netfp8 | ✓ | ✗ | ✓ |
| Novita AI | |||
| Novita AIfp8 | ✓ | ✓ | ✗ |
| DeepInfrabf16 | ✓ | ✓ | ✓ |
| DeepInfra |
Tool calling: 8 of 11 listings say yes, 3 publish no parameter list. JSON output: 6 of 11 listings say yes, 2 say no, 3 publish no parameter list. Strict schema: 5 of 11 listings say yes, 3 say no, 3 publish no parameter list.
Models people weigh against MiMo-V2.5
When we formed this view
Dates behind this page
Prices last checked 38h 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.
- 3 of 11 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.
- 3 of 11 listings do not say whether they train on prompts.
Licence and identifiers
What the licence allowsMIT License, 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
MIT License
Fully permissive: do anything with attribution. No patent grant, unlike Apache-2.0.
Identifiers
- Hugging Face
- XiaomiMiMo/MiMo-V2.5
- Architecture
- Mixture of experts
- Modality record
- text+image+audio+video->text
- Catalogue slug
- xiaomi-mimo-v2-5