MiMo-V2.6-Flash-MOPD

Run it in the Llama app: https://llama.app

Or from the terminal:

llama serve -hf ggml-org/MiMo-V2.6-Flash-MOPD-GGUF

Source models

• https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-MOPD

Notes

• The MXFP4 output keeps the routed experts at their native MXFP4 precision.
• The Q2_K output keeps the expert down projections at MXFP4, and quantizes the gate/up projections to Q2_K.
• The Q2_K_S output is smaller than Q2_K - all routed expert projections (down, gate, up) are quantized to Q2_K.
• Includes MTP sidecars (Q4_0 and Q8_0) for speculative decoding (--mtp).
• Includes a DFlash drafter sidecar (BF16 and Q8_0) for speculative decoding, converted from the dflash/ subdirectory of the source repo.
• Includes a Q8_0 mmproj for the vision and audio encoders.
• The Q2_K / Q2_K_S expert tensors are calibrated with the imatrix from https://huggingface.co/AesSedai/MiMo-V2.6-Flash-MOPD-GGUF

[!IMPORTANT] This model is automatically converted using https://github.com/ggml-org/convert