Xiaomi (MiMo)

MiMo-V2.5

DeprecatedOpen weights1M contextReasoning

Summary

MiMo-V2.5 is an open-weight language model from Xiaomi (MiMo), released on 22 Apr 2026. It accepts text, images, audio and video and generates text, with a 1M-token context window and up to 131K output tokens per response. As of 19 Nov 2026, the lowest listed API price is $0.17 per million input tokens and $0.34 per million output tokens (Novita AI), across 3 providers we track. It is scheduled for retirement on 21 Oct 2026.

Open MiMo model for multimodal coding agents and long-context automation

Last verified 19 Nov 2026. Every figure on this page links to its source.

Specifications

DeveloperXiaomi (MiMo)
Release date22 Apr 2026 [models.dev]
StatusDeprecated [models.dev]
WeightsOpen weights [Hugging Face]
Licensemit [Hugging Face]
Parameters311B [Hugging Face]
Context window1M (1,048,576 tokens) [models.dev]
Max output131K tokens [models.dev]
Knowledge cutoffDec 2024 [models.dev]
InputText, images, audio and video [models.dev]
OutputText [models.dev]
Reasoning modeYes [models.dev]
Tool callingYes [models.dev]
Retirement date21 Oct 2026 [OpenRouter]
Hugging FaceXiaomiMiMo/MiMo-V2.5 [Hugging Face]

MiMo-V2.5 API pricing by provider

USD per million tokens. Sorted by input price.

ProviderInputOutputCached inputContextSource
Novita AI
xiaomimimo/mimo-v2.5
$0.17$0.34$0.00341Mmodels.dev, 19 Nov 2026
Hugging Face Inference Providers
XiaomiMiMo/MiMo-V2.5
$0.40$2–262Kmodels.dev, 19 Nov 2026
OpenRouter
xiaomi/mimo-v2.5
$0.40$2$0.081MOpenRouter, 19 Nov 2026

Running MiMo-V2.5 locally

Estimated memory, from the parameter count and the architecture in config.json on Hugging Face. Includes about 0.8 GB of runtime overhead.

QuantizationWeightsTotal, 8K contextTotal, 32K context
Q4_K_M (4-bit)188 GB190 GB194 GB
Q8_0 (8-bit)330 GB332 GB336 GB
FP16 / BF16 (16-bit)622 GB624 GB627 GB

At 4-bit with an 8K conversation, MiMo-V2.5 needs about 190 GB, more than any single consumer graphics card; it needs a unified-memory machine, a multi-GPU setup or a split between graphics card and system RAM. Check your own hardware, or see Best computer for local AI on our sister site AIPCs.

Frequently asked questions

What is the context window of MiMo-V2.5?

MiMo-V2.5 has a context window of 1,048,576 tokens (1M) and can generate up to 131,072 tokens in a single response, according to models.dev.

How much does the MiMo-V2.5 API cost?

MiMo-V2.5 is listed from $0.17 per million input tokens and $0.34 per million output tokens (Novita AI). Prices last verified 19 Nov 2026.

Is MiMo-V2.5 open source?

MiMo-V2.5 is an open-weight model: its weights are published on Hugging Face as XiaomiMiMo/MiMo-V2.5 under the mit license. Check the license terms before commercial use.

When was MiMo-V2.5 released?

Xiaomi (MiMo) released MiMo-V2.5 on 22 Apr 2026, according to models.dev.

Which providers offer MiMo-V2.5?

We track MiMo-V2.5 on 3 providers: Novita AI, Hugging Face Inference Providers, OpenRouter.

How much VRAM does MiMo-V2.5 need?

About 190 GB at 4-bit (Q4_K_M) with an 8K-token conversation, and about 332 GB at 8-bit. That is more than any single consumer graphics card holds. These are estimates from the model's published parameter count and architecture.

Similar models from other labs

Comparable on output type, weights, release date, price and context window. Prices are the lowest listed, USD per million tokens.

ModelLabReleasedContextInput $/1MOutput $/1MWeights
DeepSeek V4 FlashDeepSeek24 Apr 20261M$0.09$0.18Open
GLM-5.3-FlashZhipu AI (Z.ai)26 Aug 20261M$0.15$0.50Open
GPT-6 Luna ProOpenAI22 Sept 20261.1M$0.10$0.50–
DeepSeek V4.1 Flash 1017DeepSeek17 Oct 20261M$0.04$0.50Open
Muse Spark 1.2 ContributorMeta21 Aug 20261M$0.10$0.20–
MiniMax-M3MiniMax1 Jun 20261M$0.28$1.10Open

Change history

  • No longer listed on vercel
  • OpenRouter price changed: input $0.12 → $0.40, output $0.24 → $2 per 1M tokens
  • OpenRouter price changed: input $0.14 → $0.12, output $0.28 → $0.24 per 1M tokens
  • vercel price changed: input $0.14 → $0.12, output $0.28 → $0.24 per 1M tokens
  • Parameter count changed from 311B to 311B
  • No longer listed on xiaomi
  • Status changed from active to deprecated
  • Retirement date changed from none to 21 Oct 2026
  • Now listed on Novita AI

Sources