MiMo-V2.5: Benchmarks, Pricing & Context Window
MiMo-V2.5 is a language model from Xiaomi, released in April 2026, with multimodal input.
MiMo-V2.5 is Xiaomi's native omnimodal sparse Mixture-of-Experts model with 310B total parameters, 15B activated parameters, and a 1M-token context window. Built on the MiMo-V2-Flash backbone, it adds dedicated vision and audio encoders
MiMo-V2.5 pricing
Providers
MiMo-V2.5 starts at $0.168 per million input tokens and $0.336 per million output tokens via Novita. See all 2 providers below with their per-token pricing, latency, throughput, and modality support.
| Provider | Input $/M | Output $/M | Max Input | Max Output | Latency s | Throughput | Quant | Input | Output |
|---|---|---|---|---|---|---|---|---|---|
 Novita | $0.168 | $0.336 | 1.0M | 131.1K | — | — | — | ||
 DeepInfra | $0.400 | $2.00 | 262.1K | 131.1K | — | — | fp8 |
MiMo-V2.5 API
API access coming soon
MiMo-V2.5 will be available through our gateway shortly.
MiMo-V2.5 examples
Recent arena outputs from MiMo-V2.5, picked from the highest-ranked matchups.
MiMo-V2.5 license
MiMo-V2.5 is released under the MIT license, which permits commercial use, has 310.8B parameters.
- License
- MIT
- Commercial use allowed
- Parameters
- 310.8B
MIT License - allows commercial use
FAQ
Common questions about MiMo-V2.5.