MiniMax, released 2026-02-12. Sparse MoE decoder, 230B total, 10B active, 192K context. Attention: GQA with QK-Norm.
Sparse omnimodal MoEXiaomi 🇨🇳 · September 21, 2026
Xiaomi MiMo-V2.6-Flash-RL (309B)
Omnimodal sparse MoE model with the MiMo-V2.5 text backbone, mixed-domain reinforcement learning, and a five-layer DFlash speculative decoder.
Key detail
Uses 256 routed experts with top-8 routing, one dense prefix layer, no shared experts, and a 128-token sliding window.Total params
309B
Active / token
15B
Context
1M
Fact sheet
- Scale
- 309B total, 15B active (4.9% active)
- Context window
- 1,048,576 tokens
- Decoder type
- Sparse omnimodal MoE
- Attention
- GQA with sliding-window/global attention
- Layer mix
- 39 sliding-window + 9 global
- KV cache / token (bf16)
- 217.5 KiB · High
- License
- MIT License
- Released
Efficiency
Parameters active per token
4.9%
15B of 309B weights run for each token.
AA Intelligence Index v5.0
Source26.7
- Knowledge Work
- 44.2
- Coding
- 22.1
- Science
- 14.4
- General
- 25.9
Architecture concepts
Sources & resources
Use the ← and → keys to step through releases.
Similar architectures
StepFun, released 2026-02-01. Sparse MoE decoder, 196B total, 11B active, 256K context. Attention: GQA with 3:1 sliding-window attention.
StepFun, released 2026-09-20. Sparse MoE decoder, 600B total, 27B active, 1M context. Attention: Gated GQA with QK-Norm and 3:1 sliding-window/sparse attention.
Thinking Machines Lab, released 2026-07-15. Sparse MoE decoder, 975B total, 41B active, 1M context. Attention: GQA with QK-Norm and learned relative-position bias in a 5:1 sliding-window/global pattern.