#10 MiMo-V2.6-Flash
#1 on LLMCheck local index (Oct 18), MIT, great dual-Spark target.
- Version
- MiMo-V2.6-Flash-MOPD
- Maker
- Xiaomi
- Release date
- 2026-09-27 (Hugging Face repo creation date unless noted)
- Architecture
- Sparse MoE (8 of 256 experts)
- Total params
- 309B
- Active params
- 15B
- Context length
- 1M
- License
- MIT
- Commercial notes
- Fully commercial
- Variants
- V2.6-Flash-MOPD
- V2.6-Flash-RL
- V2.6-Pro
- MiMo-V2.6-Distill-Qwen-9B
- Memory
- BF16/native: no published figure yet
FP8: no published figure yet
Q8: no published figure yet
Q4: no published figure yet
LLMCheck min RAM 170 GB; needs 192 GB+ Mac - Hardware
- Single Spark (128GB): No
Dual Spark (256GB): Yes (inferred from 170 GB min) unverified
Quad Spark (~512GB): Yes
24GB GPU: No
128GB Mac / Strix: No - Tokens/sec
- m3_ultra_256gb: ~59 tok/s community (LLMCheck)
- Document Q&A strengths
- 1M context
- MOPD checkpoint fixes repeated tool calls
- Embedding / reranker pairing
- no published figure yet
- Runtimes
- Ollama: Not in library (404)
llama.cpp: no published figure yet
vLLM: Yes (model card)
LM Studio: no published figure yet - Hugging Face
- https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-MOPD
DGX Spark configs it fits
- Dual Spark (256GB): Yes (inferred from 170 GB min) unverified
- Quad Spark (~512GB): Yes
Spark fit is estimated from published memory sizes, not tested on Spark hardware.
Sources
- https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-MOPD
- https://llmcheck.net/blog/state-of-open-source-local-llms-october-2026/
- https://www.benchleader.com/best/open-weights-llm
Last updated Oct 24, 2026. Fields tagged unverified lack a primary source. "No published figure yet" means we found no source, and we don't guess. Spark fit is estimated from published memory sizes, not tested on Spark hardware. Ollama listings may be cloud-only tags; check this for Kimi K3 and DeepSeek V4.1 before pulling.
The RAGShelf weekly
New local RAG tools, releases, and tested setups. One email a week, no spam.