Sulat.com
AI models
$10 off the fastest DeepSeek V4.1 Flash, Kimi K3 and GLM 5.3 from Synthetic
Deep Infra logo

Model details

MiMo-V2.5-Pro

Xiaomi MiMo V2.5 Pro sits within Xiaomi's broader MiMo model family, which is documented on the company's official MiMo updates page at mimo.mi.com. That updates page tracks the family's evolution, including a later MiMo-V2.6 series release that introduces omni-modal flagship and full-modality reasoning variants, as well as a separate mimo-v2.5-asr speech recognition model. The V2.5-Pro variant itself is named on third-party gateway listings attributed to Xiaomi authors, but it does not appear as a dedicated entry in the supplied portion of Xiaomi's official update log.

In practical terms, MiMo V2.5 Pro is described on third-party reseller pages as a text-oriented large language model that supports long context windows in the million-token range, with features such as streaming, tool calling, and structured outputs. Those gateway listings position the model as part of an open-weights family that emphasizes agent-style reasoning and developer-facing capabilities. For users evaluating MiMo V2.5 Pro, the strongest fit is general language and agent tasks within Xiaomi's MiMo ecosystem, while authoritative Xiaomi model documentation should be consulted for any production-grade deployment decisions.

Deep InfraXiaomiMiMo/MiMo-V2.5-Promimodeprecated

Quick Info

Powered by
Provider
Deep Infra
Model key
XiaomiMiMo/MiMo-V2.5-Pro
Release date
Apr 22, 2026
Last updated
Apr 22, 2026
Knowledge cutoff
2024-12
Input modalities
Output modalities
Capabilities

Cost

Input token cost
$1.00
Output token cost
$3.00

Limits

Output tokens
16,384 tokens
Context window
1,048,576 tokens

Transparent token rates

Compare MiMo-V2.5-Pro pricing

Rates are shown per one million tokens. Combined means one million input plus one million output tokens.

Browse this family

Latest news about MiMo-V2.5-Pro

Pioneer

Coverage

Xiaomi's official MiMo-V2.5-Pro model card on Hugging Face describes the model as an open-source Mixture-of-Experts language model with 1.02T total parameters and 42B active parameters, using a hybrid attention architecture that interleaves Sliding Window Attention (SWA) and Global Attention at a 6:1 ratio with a 128-t The same model card positions MiMo-V2.5-Pro as Xiaomi's most capable release to date, designed for the most demanding agentic, complex software engineering, and long-horizon tasks. It emphasizes strong instruction following and coherence over the 1M-token context, with hybrid SWA/GA reducing KV-cache storage by roughly

Videos about MiMo-V2.5-Pro

More models around MiMo-V2.5-Pro