Highlighted on the intelligence vs cost chart as compelling: a few points behind GPT 6.1 Soul Max at a fraction of the price. The creator may make a video on it.

Xiaomi MiMo
by Xiaomi
Xiaomi's MiMo family of open-weight language, multimodal, audio and speech models, with a hosted API and web demo.
Updated EU status checked Last shown
Xiaomi MiMo at a glance
- What is Xiaomi MiMo?
- Xiaomi MiMo is a model from Xiaomi. Xiaomi's MiMo family of open-weight language, multimodal, audio and speech models, with a hosted API and web demo.
- Who makes Xiaomi MiMo?
- Xiaomi MiMo is made by Xiaomi.
- Does Xiaomi MiMo work in the EU?
- Unverified. No explicit evidence found. Weights are published on Hugging Face and GitHub, but the fetched pages give no regional availability or licence terms for the hosted service. Last verified 14 Nov 2026.
- How much does Xiaomi MiMo cost?
- API pricing. Hosted API via the Xiaomi MiMo developer portal; pricing not given on fetched pages. Many open-weight models are downloadable for free.
- Where was Xiaomi MiMo shown?
- In 1 creator video by Matthew Berman. Most recently by Matthew Berman on 13 Oct 2026, at 9:50 in “Meta just dropped something insane... (Muse Wonder)”.
Sources: the vendor's own pages (linked above) and creator videos at the linked timestamps. EU status is information, not legal advice; see how we check.
MiMo is a model family from Xiaomi's MiMo team. It began with the MiMo reasoning language model (May 2025) and has expanded to MiMo-VL (vision-language), MiMo-Audio, MiMo-V2-Flash, the V2.5 series (including V2.5-Pro, an ASR model and a TTS series) and the current MiMo-V2.6 series. Xiaomi describes its work as built in public, with technical reports on reasoning, MoE reinforcement learning, sparse attention and multi-teacher on-policy distillation.
The site lists a web demo, an API developer portal and a gallery of MiMo-V2.6 outputs. Blog posts cover a long-horizon coding agent (MiMo Code), a MiMo-V2.5-Pro-UltraSpeed mode claimed to reach 1000 tokens per second on a 1T-parameter model, inference optimization with hybrid sliding-window attention, and use of V2.6-Pro in materials R&D. A creator review cited it as scoring a few points behind a top proprietary model at a fraction of the price.
Weights and code are published on Hugging Face and GitHub under the XiaomiMiMo organizations. Licensing and pricing vary by model and were not given on the fetched pages, so check each model card and the API portal.
- Model family spanning LLM, vision-language, audio and speech (ASR/TTS)
- MiMo-V2.6 series with a Pro variant aimed at agentic and coding work
- UltraSpeed mode for V2.5-Pro reporting 1000 tokens per second
- Open research with technical reports and published weights
- Web demo and developer API access
As seen on YouTube
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
Similar AI tools
Closest neighbours by meaning, not just by category.
Mimo V2.6
Model · Xiaomi
Xiaomi's MiMo-V2.6 model series, with a 1T-parameter Pro and a 311B Flash variant, plus a Mimo Desktop harness.
Xiaomi Robotics 1
Model · Xiaomi
Open-weights Vision-Language-Action robot foundation model from Xiaomi, pre-trained on 100K+ hours of UMI trajectories.
Inkling Small
Model · Thinking Machines
Open-weights multimodal MoE model (276B total, 12B active) from Thinking Machines; text, image and audio in, text out.
LFM3
Model · Liquid AI
Liquid AI's LFM3 family of small multimodal foundation models for on-device, edge and cloud use.
MiDashengLM-Gen
Model · Xiaomi
Xiaomi's LLM-driven model that generates mixed audio scenes (speech, music, SFX, ambience) from structured text captions.
MiniMax M3
Model · MiniMax
Open-weights native multimodal MoE model from MiniMax, ~428B parameters (~23B active) with 1M-token context.