Megatron MarketsWholesale
Sign inContact Us
Models Model details

Xiaomi

Xiaomi: MiMo-V2.5

xiaomi/mimo-v2.5

Enquire about this model Request a quote

MiMo-V2.5 is a native omnimodal model by Xiaomi. It delivers Pro-level agentic performance at roughly half the inference cost, while surpassing MiMo-V2-Omni in multimodal perception across image and video understanding tasks. Its 1M context window supports complete documents, extended conversations, and complex task contexts in a single pass, making it ideal for integration with agent frameworks where strong reasoning, rich perception, and cost efficiency all matter.

Input
$0.14

/ 1M tokens

Output
$0.28

/ 1M tokens

Context window
1.05M

tokens

Limits

Context

1,050,000

tokens

Max output

131,072

12% of the context window

Parameters

310,775,040,000

Model facts

Author

Xiaomi

Tokenizer

Other

Modality

Text + Image + Audio + Video -> Text

License

MIT

Created

Apr 22, 2026

Hugging Face

XiaomiMiMo/MiMo-V2.5

Default generation parameters

temperature

1

top_p

0.95

Publication status

Hugging Face: verifiedLicense: publishedParameter count: published

Modalities

Input · TextInput · AudioInput · ImageInput · VideoOutput · Text

Generation controls

Frequency penaltyInclude reasoningMax tokensPresence penaltyReasoningRepetition penaltyResponse formatSeedStopStructured outputsTemperatureTool choiceToolsTop kTop p

Capability scores

15.8
Agentic index

Better than 42% of models compared

56.8
Coding index

Better than 62% of models compared

21.8
Intelligence index

Better than 51% of models compared

Creative tasks

Category performance

UI components
Top 38%

1257 Elo

53.7% win

Data visualization
Top 33%

1250 Elo

53.6% win

Code
Top 33%

1261 Elo

53.5% win

Websites
Top 29%

1267 Elo

53.5% win

Games
Top 35%

1250 Elo

52.6% win

3D
Top 40%

1217 Elo

50.8% win

SVG
Top 50%

1176 Elo

50.0% win

ASCII art
Top 67%

1145 Elo

45.4% win