Xiaomi logo

MiMo-V2.5

Multimodal
Xiaomi

MiMo-V2.5 is Xiaomi's 310.8B-parameter open-weight multimodal MoE model released in April 2026. It leads high-resolution image understanding with a 0.89 score on HR-Bench (4k) and performs strongly on video and document benchmarks. Weights are available on Hugging Face under the MIT license.

Key Specifications

Parameters
310.8B
Context
1.0M
Release Date
April 22, 2026
Average Score
87.8%

Timeline

Key dates in the model's history
Announcement
April 22, 2026
Last Update
August 27, 2026

Technical Specifications

Parameters
310.8B
Training Tokens
-
Knowledge Cutoff
-
Family
-
Capabilities
MultimodalZeroEval

Pricing & Availability

Input (per 1M tokens)
$0.17
Output (per 1M tokens)
$0.34
Max Input Tokens
1.0M
Max Output Tokens
131.1K
Supported Features
Function CallingStructured OutputCode ExecutionWeb SearchBatch InferenceFine-tuning

Benchmark Results

Model performance metrics across various tests and benchmarks

Other Tests

Specialized benchmarks
HR-Bench (4k)
Self-reported by the model providerSelf-reported
89.0%
Video-MME
Self-reported by the model providerSelf-reported
88.0%
OmniDocBench
Self-reported by the model providerSelf-reported
87.0%
GraphWalks
Parents 1M F1Self-reported
87.0%

License & Metadata

License
mit
Announcement Date
April 22, 2026
Last Updated
August 27, 2026

Similar Models

All Models

Recommendations are based on similarity of characteristics: developer organization, multimodality, parameter size, and benchmark performance. Choose a model to compare or go to the full catalog to browse all available AI models.