XiaomiMiMo

MiMo-V2.6-Pro-RL

1. Introduction

MiMo-V2.6-Pro-RL is the flagship checkpoint of the MiMo-V2.6 series. The series is built to scale reinforcement learning toward self-improvement — scaling RL compute, environment diversity, and grader compute together, so the model keeps expanding its capability frontier through exploration and feedback. Key features include:

Native Omnimodal + Long Horizon: Text, image, video, and audio in one model; 1M tokens for long repositories, tool traces, and multi-session agent runs.
You Only RL Once: One mixed RL run across coding, general agents, visual, and cybersecurity — not separate per-domain runs. Tasks and multiple harnesses are mixed in the same batch so capabilities reinforce each other and strategies transfer to harnesses never seen in training.
Scaling RL Compute: Fully asynchronous Group Relative Policy Optimization (GRPO) on very large batches — 1,568 prompts × 16 rollouts per step, billions of tokens per update.
Groupwise Agentic Grading (Self-Improvement Loop): Binary pass/fail cannot rank passing solutions, so the reward signal itself is scaled. An agentic grader compares rollouts within each group: Groupwise Reward Synthesis (GRS) builds task-specific rubrics offline from contrasting rollouts and fuses rubric quality with test outcomes; Groupwise Advantage Redistribution (GAR) ranks passing trajectories online and moves advantage toward higher-quality solutions. Judged against the policy’s own samples, this closes a self-improvement loop and steers toward shorter paths and fewer tokens per task.
Aligned RL: Cold start from self-correction — the model reflects on and rewrites its own misaligned turns into grounded next steps. Throughout RL, environment hardening, adversarial screening, and verifier cross-checks keep the loop honest against reward hacking.
Multi-Prefix Multi-Teacher On-Policy Distillation (MOPD2): After mixed RL, MOPD2 combines autonomous student rollouts with prefix-conditioned single-turn rollouts (Teacher-Prefix and SFT-Prefix), reusing histories from teacher trajectories and SFT demonstrations so decision points train without regenerating preceding turns — extending capabilities to hard-to-verify tasks.

Model Summary

Architecture: Sparse MoE (Mixture of Experts), 1.02T total / 42B activated parameters
Context Length: 1M tokens
Modalities: Text, Image, Video, Audio
Vision Encoder: 681M-param MiMo ViT (28 layers: 24 SWA + 4 Full)
Audio Encoder: 308M AudioTokenizer + 127M audio patch encoder
Multi-Token Prediction (MTP): 5-layer speculative decoder

Figure 1: MiMo-V2.6 architecture — omni encoders, hybrid SWA backbone, and MTP blocks

Figure 1. MiMo-V2.6 architecture.

2. Downloads

Table with columns: Model, Download
Model	Download
MiMo-V2.6-Pro-RL	🤗 HuggingFace · 🤖 ModelScope (at release)
MiMo-V2.6-Flash-RL	🤗 HuggingFace · 🤖 ModelScope (at release)

3. Evaluation Results

Table with columns: Benchmark, MiMo-V2.6 Pro, MiMo-V2.6 Flash, MiMo-V2.5 Pro, Claude Opus 5, GPT-5.6 Sol, Claude Fable 5
Benchmark	MiMo-V2.6 Pro	MiMo-V2.6 Flash	MiMo-V2.5 Pro	Claude Opus 5	GPT-5.6 Sol	Claude Fable 5
Code Agent
DeepSWE v1.1	71.9	67.9	19.0	74.0	73.0

4. Model Architecture

LLM Backbone

Table with columns: Component, MiMo-V2.6-Pro-RL
Component	MiMo-V2.6-Pro-RL
Layers (Total / SWA / GA)	70 / 60 / 10
Hidden Size	6144
SWA Heads (Q/KV)	128 / 8
GA Heads (Q/KV)	128 / 8
Head Dimensions (QK / V)	192 / 128
Sliding Window Size	128
Routed Experts (Total / Activated)	384 / 8
Max Context Length	1M
MTP / Speculative Decoder

The first Transformer block uses global attention with a dense FFN. Remaining blocks interleave local SWA and GA; both use sparse MoE FFNs without shared experts.

Vision Encoder (MiMo ViT)

Table with columns: Configuration, Value
Configuration	Value
Layers (Total / SWA / GA)	28 / 24 / 4
Hidden Size	1280
Attention Heads (Q / KV)	32 / 8
Head Dimension	64
Patch Size (T × H × W)	2 × 16 × 16
Sliding Window (Left / Right)	64 / 64
Spatial Merge Size	2 × 2
Parameters	681M

Audio Encoders

AudioTokenizer encoder: 24 layers (12 SWA / 12 GA), hidden 1024, 20 RVQ codebooks, 308M parameters. Audio patch encoder: 6 layers, 127M parameters; four frames per patch (25 Hz → 6.25 Hz).

Speculative Decoder

5-layer SWA MTP drafter (DFlash-style). Predicts 7 subsequent tokens per forward pass for parallel verification.

5. Deployment

For best performance, follow the SGLang MiMo cookbook. Docker image: lmsysorg/sglang:latest.

SGLang

bash
sglang serve \
  --trust-remote-code \
  --model-path XiaomiMiMo/MiMo-V2.6-Pro-RL \
  --tp 16 \
  --dp 2 \
  --enable-dp-attention \
  --mm-enable-dp-encoder \
  --ep 16 \
  --moe-a2a-backend deepep \
  --moe-dense-tp-size 1 \
  --mem-fraction-static 0.7 \
  --max-running-requests 128 \
  --chunked-prefill-size 32768 \
  --page-size 64 \
  --swa-full-tokens-ratio 0.3 \
  --speculative-algorithm EAGLE \
  --speculative-num-steps 3 \
  --speculative-eagle-topk 1 \
  --speculative-num-draft-tokens 4 \
  --enable-multi-layer-eagle \
  --reasoning-parser mimo \
  --tool-call-parser mimo \
  --host 0.0.0.0 \
  --port 30000 \
  --nnodes 2 \
  --node-rank <node-rank> \
  --dist-init-addr <node0-ip>:20000

vLLM

Follow the vLLM MiMo-V2.5 recipe. Pre-built image: docker pull vllm/vllm-openai:mimov25-cu129.

bash
vllm serve XiaomiMiMo/MiMo-V2.6-Pro-RL \
  --tensor-parallel-size 8 \
  --trust-remote-code \
  --gpu-memory-utilization 0.95 \
  --max-model-len auto \
  --reasoning-parser mimo \
  --tool-call-parser mimo \
  --enable-auto-tool-choice \
  --generation-config vllm

Recommended sampling: temperature=1.0, top_p=0.95.

Also available in AI Studio, MiMo Code, Xiaomi MiMo Desktop, Xiaomi MiMo Open Platform API, and OpenRouter.

Citation

bibtex
@misc{mimo2026v26pro,
  title={MiMo-V2.6-Pro-RL},
  author={{Xiaomi MiMo Team}},
  year={2026},
  howpublished={\url{https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-RL}},
}

Contact

For questions or feedback, reach us at mimo@xiaomi.com or join our community:

Available on FriendliAI

Dedicated Endpoints

Run this model inference on single tenant GPU with unmatched speed and reliability at scale.

Learn more

Model Details

Model Provider

XiaomiMiMo

Model Tree

Base

this model

Input Modalities

TextAudioImageVideo

Output Modalities

Text

Supported Functionality

Dedicated Endpoints

Explore FriendliAI today

Get started Talk to an engineer

1. Introduction

Native Omnimodal + Long Horizon: Text, image, video, and audio in one model; 1M tokens for long repositories, tool traces, and multi-session agent runs.
You Only RL Once: One mixed RL run across coding, general agents, visual, and cybersecurity — not separate per-domain runs. Tasks and multiple harnesses are mixed in the same batch so capabilities reinforce each other and strategies transfer to harnesses never seen in training.
Scaling RL Compute: Fully asynchronous Group Relative Policy Optimization (GRPO) on very large batches — 1,568 prompts × 16 rollouts per step, billions of tokens per update.
Groupwise Agentic Grading (Self-Improvement Loop): Binary pass/fail cannot rank passing solutions, so the reward signal itself is scaled. An agentic grader compares rollouts within each group: Groupwise Reward Synthesis (GRS) builds task-specific rubrics offline from contrasting rollouts and fuses rubric quality with test outcomes; Groupwise Advantage Redistribution (GAR) ranks passing trajectories online and moves advantage toward higher-quality solutions. Judged against the policy’s own samples, this closes a self-improvement loop and steers toward shorter paths and fewer tokens per task.
Aligned RL: Cold start from self-correction — the model reflects on and rewrites its own misaligned turns into grounded next steps. Throughout RL, environment hardening, adversarial screening, and verifier cross-checks keep the loop honest against reward hacking.
Multi-Prefix Multi-Teacher On-Policy Distillation (MOPD2): After mixed RL, MOPD2 combines autonomous student rollouts with prefix-conditioned single-turn rollouts (Teacher-Prefix and SFT-Prefix), reusing histories from teacher trajectories and SFT demonstrations so decision points train without regenerating preceding turns — extending capabilities to hard-to-verify tasks.

Model Summary

Architecture: Sparse MoE (Mixture of Experts), 1.02T total / 42B activated parameters
Context Length: 1M tokens
Modalities: Text, Image, Video, Audio
Vision Encoder: 681M-param MiMo ViT (28 layers: 24 SWA + 4 Full)
Audio Encoder: 308M AudioTokenizer + 127M audio patch encoder
Multi-Token Prediction (MTP): 5-layer speculative decoder

Figure 1: MiMo-V2.6 architecture — omni encoders, hybrid SWA backbone, and MTP blocks

Figure 1. MiMo-V2.6 architecture.

2. Downloads

Table with columns: Model, Download
Model	Download
MiMo-V2.6-Pro-RL	🤗 HuggingFace · 🤖 ModelScope (at release)
MiMo-V2.6-Flash-RL	🤗 HuggingFace · 🤖 ModelScope (at release)

3. Evaluation Results

Table with columns: Benchmark, MiMo-V2.6 Pro, MiMo-V2.6 Flash, MiMo-V2.5 Pro, Claude Opus 5, GPT-5.6 Sol, Claude Fable 5
Benchmark	MiMo-V2.6 Pro	MiMo-V2.6 Flash	MiMo-V2.5 Pro	Claude Opus 5	GPT-5.6 Sol	Claude Fable 5
Code Agent
DeepSWE v1.1	71.9	67.9	19.0	74.0	73.0

4. Model Architecture

LLM Backbone

Table with columns: Component, MiMo-V2.6-Pro-RL
Component	MiMo-V2.6-Pro-RL
Layers (Total / SWA / GA)	70 / 60 / 10
Hidden Size	6144
SWA Heads (Q/KV)	128 / 8
GA Heads (Q/KV)	128 / 8
Head Dimensions (QK / V)	192 / 128
Sliding Window Size	128
Routed Experts (Total / Activated)	384 / 8
Max Context Length	1M
MTP / Speculative Decoder

The first Transformer block uses global attention with a dense FFN. Remaining blocks interleave local SWA and GA; both use sparse MoE FFNs without shared experts.

Vision Encoder (MiMo ViT)

Table with columns: Configuration, Value
Configuration	Value
Layers (Total / SWA / GA)	28 / 24 / 4
Hidden Size	1280
Attention Heads (Q / KV)	32 / 8
Head Dimension	64
Patch Size (T × H × W)	2 × 16 × 16
Sliding Window (Left / Right)	64 / 64
Spatial Merge Size	2 × 2
Parameters	681M

Audio Encoders

AudioTokenizer encoder: 24 layers (12 SWA / 12 GA), hidden 1024, 20 RVQ codebooks, 308M parameters. Audio patch encoder: 6 layers, 127M parameters; four frames per patch (25 Hz → 6.25 Hz).

Speculative Decoder

5-layer SWA MTP drafter (DFlash-style). Predicts 7 subsequent tokens per forward pass for parallel verification.

5. Deployment

For best performance, follow the SGLang MiMo cookbook. Docker image: lmsysorg/sglang:latest.

SGLang

bash
sglang serve \
  --trust-remote-code \
  --model-path XiaomiMiMo/MiMo-V2.6-Pro-RL \
  --tp 16 \
  --dp 2 \
  --enable-dp-attention \
  --mm-enable-dp-encoder \
  --ep 16 \
  --moe-a2a-backend deepep \
  --moe-dense-tp-size 1 \
  --mem-fraction-static 0.7 \
  --max-running-requests 128 \
  --chunked-prefill-size 32768 \
  --page-size 64 \
  --swa-full-tokens-ratio 0.3 \
  --speculative-algorithm EAGLE \
  --speculative-num-steps 3 \
  --speculative-eagle-topk 1 \
  --speculative-num-draft-tokens 4 \
  --enable-multi-layer-eagle \
  --reasoning-parser mimo \
  --tool-call-parser mimo \
  --host 0.0.0.0 \
  --port 30000 \
  --nnodes 2 \
  --node-rank <node-rank> \
  --dist-init-addr <node0-ip>:20000

vLLM

Follow the vLLM MiMo-V2.5 recipe. Pre-built image: docker pull vllm/vllm-openai:mimov25-cu129.

bash
vllm serve XiaomiMiMo/MiMo-V2.6-Pro-RL \
  --tensor-parallel-size 8 \
  --trust-remote-code \
  --gpu-memory-utilization 0.95 \
  --max-model-len auto \
  --reasoning-parser mimo \
  --tool-call-parser mimo \
  --enable-auto-tool-choice \
  --generation-config vllm

Recommended sampling: temperature=1.0, top_p=0.95.

Also available in AI Studio, MiMo Code, Xiaomi MiMo Desktop, Xiaomi MiMo Open Platform API, and OpenRouter.

Citation

bibtex
@misc{mimo2026v26pro,
  title={MiMo-V2.6-Pro-RL},
  author={{Xiaomi MiMo Team}},
  year={2026},
  howpublished={\url{https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-RL}},
}

Contact

For questions or feedback, reach us at mimo@xiaomi.com or join our community:

MiMo-V2.6-Pro-RL

README

1. Introduction

Model Summary

2. Downloads

3. Evaluation Results

4. Model Architecture

LLM Backbone

Vision Encoder (MiMo ViT)

Audio Encoders

Speculative Decoder

5. Deployment

SGLang

vLLM

Citation

Contact

Explore FriendliAI today

README

1. Introduction

Model Summary

2. Downloads

3. Evaluation Results

4. Model Architecture

LLM Backbone

Vision Encoder (MiMo ViT)

Audio Encoders

Speculative Decoder

5. Deployment

SGLang

vLLM

Citation

Contact