MiMo-V2.6-Flash
Open 309B MoE with omni-modal input at near-zero cost.
MiMo-V2.6-Flash is an open-source mixture-of-experts model with 309B total and 15B activated parameters, employing a hybrid attention mechanism for efficient long context.
- Provider
- Xiaomi
- Type
- Language model
- Released
- Sep 21, 2026
- Context window
- 1.1M tokens
- Max output
- 131K tokens
- Price
- $0.14 input / $0.28 output per 1M tokens
- Input
- text, image, audio, video
- Output
- text
- Open weights
- Yes
- License
- MIT
Best for
- High volume / low cost
- Multimodal
- Local & on-device
- Extraction & classification
Strengths
- $0.06 cost per task
- Audio & video input
More from Xiaomi
- MiMo-V2.6-Pro — Xiaomi. Xiaomi’s 1T+ open flagship — the best intelligence per dollar in the index.