XiaomiMiMoMiMoV2ForCausalLM

MiMo V2.6 Flash RL — Hardware Requirements & GPU Compatibility

ChatFunctions

MiMo V2.6 Flash RL is a 309-billion-parameter Mixture-of-Experts language model from Xiaomi's MiMo team, with about 15 billion parameters active per token, keeping generation efficient relative to its total size even though the full weight set must still fit in memory. It is built for chat and agentic use with native function-calling support, and this "Flash" checkpoint is the efficiency-oriented sibling to Xiaomi's larger MiMo-V2.6-Pro. At this scale, local inference needs server-class hardware, so most users rely on a hosted endpoint. It supports a 1,048,576 token (1M) context window. It is released under the MIT license, allowing unrestricted commercial and research use. Published on September 21, 2026, its post-training relied on large-scale, fully asynchronous reinforcement learning (GRPO).

309B params · MoE15B active1049K contextMITSep 202676.2K downloads589 likes47.9K quant downloads

Your hardware

Verdicts below are for the reference RTX 4090. Set yours to personalise the page.

Detection reads your browser's GPU name with no data sent anywhere.

MiMo V2.6 Flash RL runs on a 192 GB+ Mac or a datacenter GPU

Best local pick → Mac Pro M2 Ultra (192 GB) · Q4_K_M · 186.5 GB · ~39 tok/s

Macs from 128 GB (IQ2_M · ~54 tok/s)

Score 131.9#5 of 106 open#1 of 82 that fit ≤192 GBMIT

Not set up for it?

Reference GPU · RTX 4090 24 GB

Estimates, ±20% · how we estimate

llmrun Composite Score →

131.9

90% range 128.4–135.4
#5 of 106 open models
#1 of 82 that fit ≤192 GB

Benchmarks · tier chart ↓

Among 106 open modelsHow the Composite Score is built

Coding123.690% range 114–133
AgentsNot enough data yet
MathNot enough data yet
Science121.690% range 105–138
ReasoningNot enough data yet

Run MiMo V2.6 Flash RL

One recommended way. Every other method, OS and quant is in the step-by-step guide.Each way opens a step-by-step guide for your OS and quant.

Ways to runOther ways

each opens step-by-step

Run in cloud

Runs on B300 (288 GB) (101 GB headroom) · Q4_K_M

Generation speed: ~229 tok/s; generation speed
Cost per 1M output tokens: $8.43; per 1M output tokens

Deployon RunPodBest valuefrom $6.94/hr

$5+ bonus for new users

Compare GPUs →

Run MiMo V2.6 Flash RL

186.5 GB VRAM at 16K context · runs on one B200 192 GB

Works on macOS, Windows and Linux.

Install Ollama (Linux)
$curl -fsSL https://ollama.com/install.sh | sh
Install Ollama (macOS / Windows)
Download and run the app; it installs the ollama command.
Download Ollama ↗
Run Q4_K_M
$ollama run hf.co/vultr/MiMo-V2.6-Flash-RL-GGUF
vultr/MiMo-V2.6-Flash-RL-GGUF does not list Q4_K_M; Ollama picks a quant itself (Q4_K_M if present). Check the repo's Files tab.
Raise the context window
Type it inside the chat. For the API server instead, start it with OLLAMA_CONTEXT_LENGTH=16384 ollama serve.
$/set parameter num_ctx 16384
A larger context needs more memory than the weights alone.
Use the API
Ollama serves its API on http://localhost:11434.

Works on macOS, Windows and Linux.

Install LM Studio
Desktop app for macOS (Apple silicon), Windows and Linux.
lmstudio.ai ↗
Download the model
In the app press Ctrl/Cmd + Shift + M and paste https://huggingface.co/vultr/MiMo-V2.6-Flash-RL-GGUF, then pick Q4_K_M. Or use the CLI (run LM Studio once first):
$lms get https://huggingface.co/vultr/MiMo-V2.6-Flash-RL-GGUF
Open on Hugging Face ↗
Load and chat
Open the model loader, select the model and set Context Length to 16384. Leave GPU offload at maximum if it fits.
Start the local server (optional)
$lms server start

Works on macOS, Windows and Linux.

Install llama.cpp (macOS)
MacPorts, Nix and conda-forge also work.
$brew install llama.cpp
Install options ↗
Install llama.cpp (Windows)
$winget install llama.cpp
Install llama.cpp (Linux)
Download the prebuilt llama-*-bin-ubuntu-* archive for your GPU backend (CPU, Vulkan) from the releases page, unpack it and run the binaries from that folder. Homebrew (brew install llama.cpp) and conda-forge also work on Linux.
llama.cpp releases ↗
Start the server
Downloads the model on first run. The chat UI and an OpenAI-compatible API are at http://localhost:8080.
$llama-server -hf vultr/MiMo-V2.6-Flash-RL-GGUF -c 16384 -ngl 99 --port 8080
vultr/MiMo-V2.6-Flash-RL-GGUF does not list Q4_K_M; llama.cpp falls back to Q4_K_M or the first file.

Works on macOS, Windows and Linux.

Option A: Docker Model Runner
Docker Desktop: Settings > AI > Enable Docker Model Runner. Docker Engine on Linux: install the docker-model-plugin package.
$docker model run hf.co/vultr/MiMo-V2.6-Flash-RL-GGUF
Model Runner docs ↗
The quant tag after the repo is how Hugging Face cards write it; if your Docker version rejects it, drop the tag and pick the quant by repo.
Option B: llama.cpp server image (NVIDIA)
Needs the NVIDIA Container Toolkit. Models are cached in the llama-cache volume. Open http://localhost:8080.
$docker run --gpus all -p 8080:8080 -v llama-cache:/root/.cache ghcr.io/ggml-org/llama.cpp:server-cuda -hf vultr/MiMo-V2.6-Flash-RL-GGUF -c 16384 -ngl 99 --host 0.0.0.0 --port 8080
Image tags (cuda13, vulkan, rocm) ↗
Cache path inside the image is not in the docs; remove the -v flag if the volume complains.

Linux only.

Install vLLM
Linux only; on Windows run it inside WSL2. NVIDIA GPU shown; AMD ROCm, Intel and CPU builds have their own install pages.
$uv venv --python 3.12 --seed
$source .venv/bin/activate
$uv pip install vllm --torch-backend=auto
vLLM installation ↗
Serve the model
OpenAI-compatible API at http://localhost:8000/v1. Splits the model over 8 GPUs of one server (tensor parallel).
$vllm serve XiaomiMiMo/MiMo-V2.6-Flash-RL --max-model-len 16384 --tensor-parallel-size 8
Runs the original weights, not the Q4_K_M GGUF; needs ~680 GB of GPU memory.

macOS only.

MLX supports only macOS, so follow these steps for that setup.

Install mlx-lm
$pip install mlx-lm
Chat in the terminal
$mlx_lm.chat --model inferencerlabs/MiMo-V2.6-Flash-RL-MLX-Q9
Downloads the model on first run. The MLX repo has its own quantisation; it may not match the quant selected above.
Or start a local server
Listens on http://localhost:8080.
$mlx_lm.server --model inferencerlabs/MiMo-V2.6-Flash-RL-MLX-Q9

Works on macOS, Windows and Linux.

Install Jan
jan.ai ↗
Find the model
Open the Hub tab, paste this model ID in Search, and choose Q4_K_M if offered.
$vultr/MiMo-V2.6-Flash-RL-GGUF
Use the local API (optional)
Settings > Local API Server > Start Server. It listens on http://127.0.0.1:1337.

Rent GPUs hourly to run Q4_K_M right there. Pricing comes from partner marketplaces and varies.

Available GPU rentals for Q4_K_M, prioritized by value
GPU	Speed	$/1M tokens	Rent
B300 (288 GB)CheapestCheapest101 GB free	~229 tok/s	$8.43	Deploy on RunPodfrom $6.94/hr (opens in new tab)	288 GB · 101 GB free · ~229 tok/s · $8.43 / 1M tokens

Compare GPUs →

How Much VRAM Does MiMo V2.6 Flash RL Need?

Select a quantization level; the listed hardware adheres to it.

VRAM	+32K	+128K	File	Quality	Runs on
93.1 GB	94.6 GB	99.4 GB	98.7 GB	2-bit · i-quant extra small — Importance-weighted 2-bit, extra small	Runs on RTX PRO 6000 Blackwell Max-Q Workstation Edition 96 GB
104.7 GB	106.2 GB	111 GB	107.8 GB	2-bit · i-quant medium — Importance-weighted 2-bit, medium	Runs on Instinct MI250X 128 GB
131.7 GB	133.2 GB	138.1 GB	126.4 GB	2-bit · K-quant — 2-bit quantization with K-quant improvements	Runs on H200 NVL 141 GB
131.7 GB	133.2 GB	138.1 GB	146.8 GB	3-bit · i-quant small — Importance-weighted 3-bit, small	Runs on H200 NVL 141 GB
139.4 GB	141 GB	145.8 GB	148.6 GB	3-bit · i-quant medium — Importance-weighted 3-bit, medium	Runs on H200 NVL 141 GB
151.0 GB	152.6 GB	157.4 GB	~150.6 GB	3-bit medium — 3-bit medium quantization	Runs on B200 192 GB
154.9 GB	156.4 GB	161.2 GB	~154.5 GB	4-bit · legacy — 4-bit legacy quantization	Runs on B200 192 GB
185.8 GB	187.3 GB	192.1 GB	~185.4 GB	4-bit medium — 4-bit medium quantization — most popular sweet spot	Runs on B200 192 GB
220.6 GB	222.1 GB	226.9 GB	~220.2 GB	5-bit medium — 5-bit medium quantization — good quality/size tradeoff	Runs on Instinct MI325X 256 GB
255.3 GB	256.8 GB	261.7 GB	~254.9 GB	6-bit, very good — 6-bit quantization, very good quality	Runs on Instinct MI350X 288 GB
309.4 GB	310.9 GB	315.7 GB	~309 GB	8-bit, near-lossless — 8-bit quantization, near-lossless	Runs on 5× H100 SXM 80 GB
618.4 GB	619.9 GB	624.7 GB	~618 GB	16-bit · bfloat — Brain floating point 16 — preferred for training	Runs on 5× H200 SXM 141 GB

est.calculated, no published file yet

Selected Q4_K_M

4-bit medium quantization

186.5GBat 16K context

Weights: 185.4 GB
KV cache: 0.8 GB
Overhead: 0.3 GB

Server-class: ~186.5 GB at 16K context · one B200 192 GB

Or .

Which Hardware Can Run MiMo V2.6 Flash RL?

Q4_K_M · 185.8 GB

MiMo V2.6 Flash RL (Q4_K_M) requires 185.8 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 242+ GB is recommended. Using the full 1049K context window can add up to 51.4 GB, bringing total usage to 237.2 GB. No consumer GPU has enough memory. 4 devices with unified memory can run MiMo V2.6 Flash RL, including Mac Studio (M3 Ultra, 512GB), Mac Studio (M3 Ultra, 256GB).

By VRAM tier · typical GPUs and Macs at Q4_K_M

Consumer GPUs (up to 32 GB) and Macs up to 96 GB: too small for this model (needs ≥ 93.8 GB)8 GB tier →12 GB tier →16 GB tier →24 GB tier →Macs →

How good is MiMo V2.6 Flash RL?

The llmrun Score merges 4 public benchmark boards into a single figure, where 100 represents the average for open models.

Score vs VRAM in the ≤192 GB tier

MiMo V2.6 Flash RL
Other models
Best Score for its VRAM
x: VRAM (GB, log scale) · y: llmrun Score

#1 of 82 that fit ≤192 GB

Each dot is a model at the best quant that fits ≤192 GB. The line steps up to the best Score available at each amount of VRAM; a dot below it is beaten by a model that needs no more memory.

Full ≤192 GB ranking →

Benchmarks

LMArena Text
LMArena · Oct 2026Arena
1447.6
#16/188 open
Arena
LMArena WebDev
LMArena · Oct 2026Arena
1637.2
#3/55 open
Arena
SciCode
Epoch AI · Sep 2026Lab-reported
51.3%
#12/72 open
Lab-reported
ALE-Bench
Epoch AI · Sep 2026Lab-reported
1244.2
#8/48 open
Lab-reported
ProofBench
Epoch AI · Sep 2026Lab-reported
63.0%
#5/31 open
Lab-reported
CritPt
Epoch AI · Sep 2026Lab-reported
12.0%
#14/84 open
Lab-reported

All 6 MiMo V2.6 Flash RL benchmarks

In the ≤192 GB class

82 models fit ≤192 GB

Same Score, only models that fit ≤192 GB at their best quant: the ranking a ≤192 GB owner actually chooses from.

#1
MiMo V2.6 Flash RLXiaomi
131.9
~229 tok/s · 186.5 GB
#2
Step 5 PreviewStepFun
128.0
~151 tok/s · 191.4 GB
#3
DeepSeek V4 Flash 0731DeepSeek
127.6
~181 tok/s · 183.0 GB
#4
GLM 5.3 FlashZ.ai
124.1
~228 tok/s · 185.1 GB
#5
Qwen3.8 27BAlibaba
122.0
~299 tok/s · 17.9 GB
#6
Inkling SmallThinking Machines
119.6
~399 tok/s · 162.7 GB

Full ≤192 GB ranking →B200 page →

Alternatives to MiMo V2.6 Flash RL

MiMo family · 8 models →

Official models with an llmrun Score, each answering one question; sizes are each model's default quant (usually Q4_K_M) with a 16K-token context.

Runner-up at this size
DeepSeek V4 Flash 0731
DeepSeek
- 4.3 lower Score
- 3.5 GB less VRAM
Score 127.6 · 183 GB
Compare with MiMo V2.6 Flash RL Open DeepSeek V4 Flash 0731
Other vendor
GLM 5.3 Flash
Z.ai
- 7.8 lower Score
- 18.7 GB more VRAM
Score 124.1 · 205.2 GB
Compare with MiMo V2.6 Flash RL Open GLM 5.3 Flash

Similar models

Model	Score	VRAM	tok/s	Why it is similar
Qwen3.8 27BAlibaba	122.0	17.9 GB	~38	similar score
Inkling SmallThinking Machines	119.6	162.7 GB	—needs > 24 GB	similar score
Beam 1.0 501B A23Breflection	119.5	304 GB	—needs > 24 GB	similar score
DeepSeek V4 FlashDeepSeek	115.3	175.1 GB	—needs > 24 GB	same size
GLM 4.7Z.ai	109.1	217.9 GB	—needs > 24 GB	same size
Qwen3 235B A22B Thinking 2507Alibaba	108.4	142.9 GB	—needs > 24 GB	same size
GLM 4.6Z.ai	103.1	216.9 GB	—needs > 24 GB	same size
GLM 4.5Z.ai	100.9	217.9 GB	—needs > 24 GB	same size
ERNIE 5.5 Flash BaseBaidu	—	573 GB	—needs > 24 GB	same size
IQuest Q1IQuestLab	—	195.4 GB	—needs > 24 GB	same size

Fine-tunes & variants(1)community builds of MiMo V2.6 Flash RL

MiMo V2.6 Flash RL UNCENSORED

Specifications

Publisher: Xiaomi
Family: MiMo
Parameters: 309B
Active parameters: 15B per token
Architecture: MiMoV2ForCausalLM
Context Length: 1,048,576 tokens
Vocabulary Size: 152,576
Release Date: 2026-09-21
License: MIT

Resources

Frequently Asked Questions

How much VRAM does MiMo V2.6 Flash RL need?

MiMo V2.6 Flash RL requires 185.8 GB of VRAM at Q4_K_M, or 618.4 GB at BF16. Full 1049K context adds up to 51.4 GB (237.2 GB total).

VRAM = Weights + KV Cache + Overhead

Weights = 309B × 4.8 bits ÷ 8 = 185.4 GB

KV Cache + Overhead ≈ 0.4 GB (at 2K context + ~0.3 GB framework)

Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.

KV Cache + Overhead ≈ 51.8 GB (at full 1049K context)

VRAM usage by quantization

Q4_K_M

185.8 GB

Q4_K_M + full context

237.2 GB

Learn more about VRAM estimation →

Can NVIDIA GeForce RTX 5090 run MiMo V2.6 Flash RL?

No — MiMo V2.6 Flash RL requires at least 93.1 GB at IQ2_XS, which exceeds the NVIDIA GeForce RTX 5090's 32 GB of VRAM.

What's the best quantization for MiMo V2.6 Flash RL?

For MiMo V2.6 Flash RL, Q4_K_M (185.8 GB) offers the best balance of quality and VRAM usage. Q5_K_M (220.6 GB) provides better quality if you have the VRAM. The smallest option is IQ2_XS at 93.1 GB.

VRAM requirement by quantization

IQ2_XS

93.1 GB~57%

IQ3_S

131.7 GB~75%

Q4_0

154.9 GB~85%

Q4_K_M ★

185.8 GB~89%

Q5_K_M

220.6 GB~92%

BF16

618.4 GB~100%

★ Recommended — best balance of quality and VRAM usage.

Learn more about quantization →

Can I run MiMo V2.6 Flash RL on a Mac?

Yes — Mac Pro M2 Ultra (128 GB) and 14 other Macs can run MiMo V2.6 Flash RL, the smallest at IQ2_M. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 185.8 GB of usable unified memory (RAM minus macOS overhead), from Mac Pro M2 Ultra (192 GB).

Can I run MiMo V2.6 Flash RL locally?

Yes — MiMo V2.6 Flash RL can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.

How fast is MiMo V2.6 Flash RL?

At Q4_K_M, MiMo V2.6 Flash RL can reach ~87 tok/s on AMD Instinct MI350X. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.

tok/s = 1000 ÷ (active GB ÷ (bandwidth GB/s × efficiency) × 1000 + layers × routing ms)

Mixture-of-Experts: just the active experts get read for each token, along with a fixed routing cost per layer.

Example: NVIDIA B200 → 9.0 GB active ÷ (8000 × 0.65) = 1.73 ms, plus 48 layers × 0.055 ms = 2.64 ms, so 1000 ÷ 4.37 ms = ~229 tok/s

Estimated speed at Q4_K_M (185.8 GB)

NVIDIA B200

~229 tok/s

NVIDIA B300

~229 tok/s

AMD Instinct MI350X

~87 tok/s

Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.

Learn more about tok/s estimation →

What's the download size of MiMo V2.6 Flash RL?

At Q4_K_M, the download is about 185.40 GB. The full-precision BF16 version is 618.00 GB. The smallest option (IQ2_XS) is 92.70 GB.

Which GPUs can run MiMo V2.6 Flash RL?

No single consumer GPU has enough VRAM to run MiMo V2.6 Flash RL at Q4_K_M (185.8 GB). Multi-GPU or professional hardware is required.

Which devices can run MiMo V2.6 Flash RL?

4 devices with unified memory can run MiMo V2.6 Flash RL at Q4_K_M (185.8 GB), including Mac Pro M2 Ultra (192 GB), Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB), Mac Studio M2 Ultra (192 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.

MiMo V2.6 Flash RL — Hardware Requirements & GPU Compatibility

MiMo V2.6 Flash RL runs on a 192 GB+ Mac or a datacenter GPU

llmrun Composite Score →

Run MiMo V2.6 Flash RL

Recommended

Ways to runOther ways

Run in cloud

How Much VRAM Does MiMo V2.6 Flash RL Need?

Selected Q4_K_M

Which Hardware Can Run MiMo V2.6 Flash RL?

How good is MiMo V2.6 Flash RL?

Score vs VRAM in the ≤192 GB tier

Benchmarks

In the ≤192 GB class

Alternatives to MiMo V2.6 Flash RL

DeepSeek V4 Flash 0731

GLM 5.3 Flash

Similar models

Specifications

Resources

Frequently Asked Questions