AI provider / Xiaomi
Xiaomi
A fast-rising Chinese provider centered on MiMo, with a recent push from text reasoning into vision, audio, speech and agent-ready multimodal releases.
Xiaomi now reads as more than a device company experimenting with models. The MiMo site already shows a layered first-party stack: the original MiMo line, MiMo-VL for vision, MiMo-Audio for speech, and the V2 and V2.5 branches that split into Flash, Pro, Omni, ASR and newer distilled variants.
What makes the provider interesting is the bridge between consumer hardware and model deployment. MiMo is not only adding benchmark-style capability, but moving toward the kinds of language, voice and multimodal building blocks that fit assistants, phones, wearables and broader device-native products.
Model catalog
Every release in this provider’s model line.
Each row is a distinct model release — its date, its type and the traits it is known for. Open a source to dig into any of them.
| Model | Released | Type | Key traits | Source |
|---|---|---|---|---|
| MiMo-V2.6 | Sep 2026 | Omnimodal |
| |
| MiMo-V2.5-DFlash | Jul 2026 | Cost optimized |
| |
| MiMo-V2.5-Pro | Apr 2026 | Flagship |
| |
| MiMo-V2.5-ASR | Apr 2026 | Audio |
| |
| MiMo-V2.5 | Apr 2026 | Version update |
| |
| MiMo-V2-Omni | Mar 2026 | Multimodal |
| |
| MiMo-V2-Pro | Mar 2026 | Flagship |
| |
| MiMo-V2-Flash | Jan 2026 | Fast tier |
| |
| MiMo-Audio | Sep 2025 | Audio |
| |
| MiMo-VL | Jun 2025 | Multimodal |
| |
| MiMo | May 2025 | Debut |
|
Timeline
A practical timeline of model releases and company milestones.
This page tracks the moments that matter most: major model releases, version jumps and company moves that changed how this provider shows up in the AI market.
MiMo-V2.6 goes fully omnimodal at 1 trillion parameters.
Five months after V2.5, the open-weight Pro variant doubles the Intelligence Index score (26 to 46), natively handling text, image, video and audio with a 1M-token context.
MiMo V2.5 DFlash shows Xiaomi iterating the line for faster public checkpoints.
The newest Hugging Face release suggests Xiaomi is not only scaling the flagship branch, but also optimizing the MiMo family for lighter deployment tiers.
MiMo V2.5 Pro becomes Xiaomi's current top public tier.
The Pro branch is where Xiaomi concentrates its strongest public positioning around agent workloads, context length and top-end performance.
MiMo V2.5 resets the main line around a newer base.
The 2.5 branch becomes the new reference point for Xiaomi's public MiMo family and sets up the later Pro, ASR and distilled variants.
MiMo V2 Pro and Omni widen the stack from text to agentic multimodality.
Xiaomi pairs a long-context Pro tier with an Omni branch, making the MiMo family look much closer to a deployable assistant platform.
MiMo-Audio makes speech a first-class MiMo modality.
The release links Xiaomi's model roadmap more directly to voice interfaces and device-native assistant use cases.
MiMo-VL pushes Xiaomi into vision-language work.
Only weeks after the original MiMo debut, Xiaomi adds a dedicated visual branch and starts framing MiMo as a broader multimodal stack.
MiMo opens Xiaomi's public foundation-model line.
The first MiMo release gives Xiaomi a clear language-model identity and starts the company's shift from device ecosystem player to first-party model producer.