MiMo Models by Xiaomi

XiaomiProprietary
Xiaomi releases · 10 in the last 12 months · this family litChangelog →
7 models2026Up to 1M ctxFrom $0.140/1M input

Last refreshed 2026-10-20. Next refresh: weekly.

Details

ResearcherXiaomi
LicenseProprietary
Commercial useRestricted
Models7
Released2026
Max context1M

Capabilities

Vision3 of 7 models
Multimodal5 of 7 models
Reasoning3 of 7 models
JSON / Tool use5 of 7 models
Structured Outputs3 of 7 models

About

MiMo is Xiaomi’s foundation-model family for reasoning, coding, agents, and multimodal work, documented on Xiaomi’s MiMo developer site. The line spans text and omni-modal releases (for example MiMo V2 Flash/Pro/Omni and later V2.5 and V2.6 Pro/Flash variants), plus related speech recognition and text-to-speech models. Xiaomi positions larger Pro-class models for long-horizon and high-stakes agent tasks and Flash-class models for higher-frequency, lower-cost serving, with hybrid attention designs and very long context windows called out on the model-update notes.

Decision facts

Best fit
text to speechagenticlong context
Capability starting point
MiMo-V2.6-Pro with 1M context and reasoning, JSON / Tool use, and multimodal inputs
Lowest tracked input
MiMo-V2.6-Flash · $0.140/1M · OpenRouter
Closest related family
MiMo V2

Current Variants

Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.

7 in view

Use when the workload needs 1M context, reasoning, and JSON / Tool use.

2026-091M contextreasoningJSON / Tool use

Use when the workload needs 1M context, reasoning, and JSON / Tool use.

2026-091M contextreasoningJSON / Tool use

Use when the workload needs text to speech, multimodal inputs, and audio.

2026-04text to speechmultimodal inputsaudio
MiMo-V2.5Current

Use when the workload needs agentic, 1M context, and reasoning.

2026-04agentic1M contextreasoning

Use when the workload needs code, 1M context, and 1000B parameters.

2026-04code1M context1000B parameters

Use when the workload needs 1M context, 1000B parameters, and JSON / Tool use.

2026-031M context1000B parametersJSON / Tool use

Use when the workload needs 200K context, multimodal inputs, and audio.

2026-02200K contextmultimodal inputsaudio

Release Timeline

4 release groups
2026-09
2 current
MiMo-V2.6-Flash
1M contextreasoningJSON / Tool use
Current
MiMo-V2.6-Pro
1M contextreasoningJSON / Tool use
Current
2026-04
3 current
MiMo-V2.5
agentic1M contextreasoning
Current
MiMo-V2.5-Pro
code1M context1000B parameters
Current
MiMo-V2.5-TTS-Series
text to speechmultimodal inputsaudio
Current
2026-03
1 current
Xiaomi MiMo-V2-Pro
1M context1000B parametersJSON / Tool use
Current
2026-02
1 current
Xiaomi MiMo-V2-Omni
200K contextmultimodal inputsaudio
Current

Specifications(7 models)

MiMo model specifications comparison
ModelReleasedContextParametersVisionMultimodalReasoningJSON / Tool useStructured Outputs
MiMo-V2.6-Pro2026-091M1.02T total / 42B active (MoE)YesYesYesYesNo
MiMo-V2.6-Flash2026-091M309B total / 15B active (MoE)YesYesYesYesYes
MiMo-V2.5-TTS-Series2026-04——NoYesNoNoNo
MiMo-V2.52026-041M—YesYesYesYesYes
MiMo-V2.5-Pro2026-041M1TNoNoNoYesYes
Xiaomi MiMo-V2-Pro2026-031M1TNoNoNoYesNo
Xiaomi MiMo-V2-Omni2026-02200K—NoYesNoNoNo

Available From(4 providers)

Pricing

MiMo model pricing by provider
ModelProviderInput / 1MOutput / 1MType
MiMo-V2.6-FlashOpenRouter$0.14$0.28Serverless
MiMo-V2.5OpenRouter$0.14$0.28Serverless
MiMo-V2.5Xiaomi$0.14$0.28Serverless
MiMo-V2.6-FlashXiaomi$0.14$0.28Serverless
MiMo-V2.6-FlashFlexAI$0.14$0.28Serverless
MiMo-V2.5-ProOpenRouter$0.435$0.87Serverless
MiMo-V2.6-ProOpenRouter$0.435$0.87Serverless
MiMo-V2.5-ProXiaomi$0.435$0.87Serverless
MiMo-V2.6-ProXiaomi$0.435$0.87Serverless
MiMo-V2.5-ProNovita AI$2$6Serverless

Popular comparisons in this family