Model tracker / Xiaomi
MiMo-V2.6 Pro
Open-weight 1T MoE model (42B active) for agentic coding and tool use; multimodal input
MiMo-V2.6 Pro is an everyday open-weight model from Xiaomi, released 21 September 2026. At DeepInfra it costs $0.43 per million input tokens and $0.87 per million output tokens, with a 1M-token context window. 2 providers serve it; the cheapest is DeepInfra at $0.43 input and $0.87 output.
- Made by
- Xiaomi (MiMo)
- Released
- 2026-09-21
- Status
- Generally available
- Tier
- Everyday
- Weights
- Open
- Memory to run
- About 680 GB (MXFP4)
- Context window
- 1,048,576 tokens
- Input
- text, image, video, audio
- API id
XiaomiMiMo/MiMo-V2.6-Pro
MiMo-V2.6 Pro price at 2 providers
US dollars per million tokens. Blended is 3 input tokens for every output token, one number to compare on. The cheapest blended price is highlighted.
| Provider | Input | Output | Cached input | Blended | Batch | Checked |
|---|---|---|---|---|---|---|
| DeepInfra | $0.43 | $0.87 | $0.004 | $0.54 | – | 2026-10-27 |
| Novita AI · xiaomimimo/mimo-v2.6-pro | $0.435 | $0.87 | $0.004 | $0.544 | – | 2026-10-27 |
- DeepInfra: DeepInfra currently shows 10% off, with no end date: $0.387 in, $0.783 out, $0.0032 cached. Served in FP8.
- Novita AI: Served in FP8.
Can you run MiMo-V2.6 Pro yourself?
Yes. Xiaomi publishes the weights on Hugging Face as XiaomiMiMo/MiMo-V2.6-Pro-MOPD, under MIT: 1.02T (MoE, 42B active) parameters, 566 GB in MXFP4.
Running it takes about 680 GB of GPU memory: more than an eight-GPU server with 80 GB cards holds (640 GB): bigger GPUs or several servers. That's the weight files plus 20% for working memory with short inputs; long contexts need more, and fitting in memory doesn't make it fast.
MOPD upgrade of the MiMo-V2.6-Pro-RL checkpoint (same size)
Models like MiMo-V2.6 Pro
Other everyday models closest in price, at each maker's own price or the cheapest provider.
| Model | Input | Output | Context | Weights |
|---|---|---|---|---|
| MiMo-V2.6 Pro | $0.43 | $0.87 | 1M | Open |
| MiniMax M3 MiniMax | $0.30 | $1.20 | 1M | Open |
| Beam Reflection | $0.25 | $1.00 | 1M | Open |
| Qwen3.7 Plus Alibaba | $0.40 | $1.60 | 1M | Closed |
| Mistral Large 3 Mistral AI | $0.50 | $1.50 | 256K | Open |
| NVIDIA Nemotron 3 Ultra NVIDIA | $0.50 | $2.20 | 262K | Open |
Compare them all in the model tracker →
MiMo-V2.6 Pro FAQ
How much does MiMo-V2.6 Pro cost?
MiMo-V2.6 Pro costs $0.43 per million input tokens and $0.87 per million output tokens at DeepInfra, and $0.004 per million cached input tokens. Prices are list prices in US dollars, last checked 27 October 2026.
What is MiMo-V2.6 Pro's context window?
MiMo-V2.6 Pro takes up to 1,048,576 tokens in one request.
Is MiMo-V2.6 Pro open weight?
Yes. Xiaomi publishes the weights, so you can run MiMo-V2.6 Pro yourself or at any host that serves it.
Can I run MiMo-V2.6 Pro locally?
MiMo-V2.6 Pro's weights are 566 GB in MXFP4, so it needs about 680 GB of GPU memory: more than an eight-GPU server with 80 GB cards holds (640 GB): bigger GPUs or several servers.
Where can I use MiMo-V2.6 Pro?
MiMo-V2.6 Pro is available from DeepInfra and Novita AI. At DeepInfra, call it as XiaomiMiMo/MiMo-V2.6-Pro.