Instructions to use inferencerlabs/MiMo-V2.5-LM-MLX-Q9 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use inferencerlabs/MiMo-V2.5-LM-MLX-Q9 with MLX:

# Make sure mlx-lm is installed
# pip install --upgrade mlx-lm

# Generate text with mlx-lm
from mlx_lm import load, generate

model, tokenizer = load("inferencerlabs/MiMo-V2.5-LM-MLX-Q9")

prompt = "Write a story about Einstein"
messages = [{"role": "user", "content": prompt}]
prompt = tokenizer.apply_chat_template(
    messages, add_generation_prompt=True
)

text = generate(model, tokenizer, prompt=prompt, verbose=True)

Notebooks
Google Colab
Kaggle
Local Apps Settings
LM Studio

How to use inferencerlabs/MiMo-V2.5-LM-MLX-Q9 with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "inferencerlabs/MiMo-V2.5-LM-MLX-Q9"

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "mlx-lm": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "inferencerlabs/MiMo-V2.5-LM-MLX-Q9"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use inferencerlabs/MiMo-V2.5-LM-MLX-Q9 with Hermes Agent:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "inferencerlabs/MiMo-V2.5-LM-MLX-Q9"

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default inferencerlabs/MiMo-V2.5-LM-MLX-Q9

Run Hermes

hermes

OpenClaw new

How to use inferencerlabs/MiMo-V2.5-LM-MLX-Q9 with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "inferencerlabs/MiMo-V2.5-LM-MLX-Q9"

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "inferencerlabs/MiMo-V2.5-LM-MLX-Q9" \
  --custom-provider-id mlx-lm \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

MLX LM

How to use inferencerlabs/MiMo-V2.5-LM-MLX-Q9 with MLX LM:

Generate or start a chat session

# Install MLX LM
uv tool install mlx-lm
# Interactive chat REPL
mlx_lm.chat --model "inferencerlabs/MiMo-V2.5-LM-MLX-Q9"

Run an OpenAI-compatible server

# Install MLX LM
uv tool install mlx-lm
# Start the server
mlx_lm.server --model "inferencerlabs/MiMo-V2.5-LM-MLX-Q9"
# Calling the OpenAI-compatible server with curl
curl -X POST "http://localhost:8000/v1/chat/completions" \
   -H "Content-Type: application/json" \
   --data '{
     "model": "inferencerlabs/MiMo-V2.5-LM-MLX-Q9",
     "messages": [
       {"role": "user", "content": "Hello"}
     ]
   }'

MiMo-V2.5-LM-MLX-Q9 / README.md

inferencerlabs

Upload model file

0256227 verified 3 months ago

preview code

Raw

History Blame

2.64 kB

	---
	language: en
	library_name: mlx
	tags:
	- quantized
	- mlx
	base_model: XiaomiMiMo/MiMo-V2.5
	base_model_relation: quantized
	pipeline_tag: text-generation
	---
	# NOTICE
	<marquee direction="left" width="400">
	<h1 style="font-size: 60px; color: red;">CURRENTLY UPLOADING...     </h1>
	</marquee>

	See MiMo-V2.5 MLX in action - [demonstration videos](https://youtube.com/xcreate)

	#### Tested on a M3 Ultra 512GB RAM using [Inferencer app v1.11.1](https://inferencer.com)
	- Text inference: ~28 tokens/s @ 2000 tokens (debug build)
	- Omnimodal inference: Not included in this language model (LM) only version
	- Memory usage: ~323 GiB

	![Screenshot](https://cdn-uploads.huggingface.co/production/uploads/688479d616f1ec82fa645019/NSFLx0coOpGbtelQHIRbN.jpeg)

	<p style="margin-bottom:0px;">
	<strong>Q9 typically achieves near lossless accuracy in our coding test</strong>
	</p>
	<table style="border-collapse: collapse; text-align:center; margin-top:10px; margin-bottom:0px;">
	<thead>
	<tr><th>Quantization (bpw)</th><th>Perplexity</th><th>Token Accuracy</th><th>Missed Divergence</th></tr>
	</thead>
	<tbody>
	<tr><td><strong>Q3.5</strong></td><td>197.0</td><td>44.05%</td><td>72.00%</td></tr>
	<tr><td><strong>Q4.5</strong></td><td>1.35937</td><td>89.75%</td><td>28.98%</td></tr>
	<tr><td><strong>Q5.5</strong></td><td>1.24218</td><td>94.60%</td><td>17.55%</td></tr>
	<tr><td><strong>Q6.5</strong></td><td>1.21875</td><td>96.85%</td><td>16.03%</td></tr>
	<tr><td><strong>Q8.5</strong></td><td>1.21875</td><td>97.65%</td><td>9.92%</td></tr>
	<tr><td><strong>Q9</strong></td><td>1.21093</td><td>97.80%</td><td>9.60%</td></tr>
	<tr><td><strong>Base</strong></td><td>1.20312</td><td>100.0%</td><td>0.000%</td></tr>
	</tbody>
	</table>

	- Perplexity: Measures the confidence for predicting base tokens (lower is better)
	- Token Accuracy: The percentage of correctly generated base tokens
	- Missed Divergence: Measures severity of misses; how much the token was missed by

	##### Quantized with a modified version of [MLX](https://github.com/ml-explore/mlx)
	##### For more details see [demonstration videos](https://youtube.com/xcreate) or visit [XiaomiMiMo/MiMo-V2.5](https://huggingface.co/XiaomiMiMo/MiMo-V2.5).

	## Disclaimer

	We are not the creator, originator, or owner of any model listed. Each model is created and provided by third parties. Models may not always be accurate or contextually appropriate. You are responsible for verifying the information before making important decisions. We are not liable for any damages, losses, or issues arising from its use, including data loss or inaccuracies in AI-generated content.