Instructions to use Anotos/LFM2-2.6B-MLX-4bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Anotos/LFM2-2.6B-MLX-4bit 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("Anotos/LFM2-2.6B-MLX-4bit")

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 Anotos/LFM2-2.6B-MLX-4bit with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "Anotos/LFM2-2.6B-MLX-4bit"

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "mlx-lm": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "Anotos/LFM2-2.6B-MLX-4bit"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

MLX LM

How to use Anotos/LFM2-2.6B-MLX-4bit with MLX LM:

Generate or start a chat session

# Install MLX LM
uv tool install mlx-lm
# Interactive chat REPL
mlx_lm.chat --model "Anotos/LFM2-2.6B-MLX-4bit"

Run an OpenAI-compatible server

# Install MLX LM
uv tool install mlx-lm
# Start the server
mlx_lm.server --model "Anotos/LFM2-2.6B-MLX-4bit"
# Calling the OpenAI-compatible server with curl
curl -X POST "http://localhost:8000/v1/chat/completions" \
   -H "Content-Type: application/json" \
   --data '{
     "model": "Anotos/LFM2-2.6B-MLX-4bit",
     "messages": [
       {"role": "user", "content": "Hello"}
     ]
   }'

Hermes Agent

How to use Anotos/LFM2-2.6B-MLX-4bit 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 "Anotos/LFM2-2.6B-MLX-4bit"

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 Anotos/LFM2-2.6B-MLX-4bit

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use Anotos/LFM2-2.6B-MLX-4bit with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "Anotos/LFM2-2.6B-MLX-4bit"

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 "Anotos/LFM2-2.6B-MLX-4bit" \
  --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"

LFM2 2.6B, MLX 4-bit

An MLX 4-bit conversion of LiquidAI/LFM2-2.6B (commit 36ed799f70), for Apple Silicon. Small enough for iPhones with 6 GB of memory.

What was changed

Converted and quantized with mlx-lm 0.32.0:

mlx_lm.convert --hf-path LiquidAI/LFM2-2.6B -q --q-bits 4

4-bit weights (4.5 bits per weight overall, group size 64). No other changes. The file is 1.45 GB and the model peaks at about 1.9 GB of memory on an M4 Mac.

model.safetensors SHA-256: 7cebcdbbcd0ff753616a22fb8dfe98f9cdc11a256430c5b87fbec09342e2cf46

Use

mlx_lm.generate --model Anotos/LFM2-2.6B-MLX-4bit --prompt "..."

Related models

The models Erato uses, all MLX 4-bit:

Model	For	Notes
Anotos/LFM2-8B-A1B-MLX-4bit	12 GB iPhones, 16 GB Macs	Built into Erato
Anotos/Gemma-4-E4B-it-Text-MLX-4bit	8 GB iPhones and Macs	Built into Erato
Anotos/LFM2-2.6B-MLX-4bit	Devices under 8 GB	Built into Erato
Anotos/LFM2-8B-A1B-Abliterated-MLX-4bit	12 GB iPhones, 16 GB Macs	Uncensored. Not built in; add it yourself
Anotos/Gemma-4-E4B-Heretic-MLX-4bit	8 GB iPhones and Macs	Uncensored. Not built in; add it yourself

The uncensored models have had their guardrails removed and can generate dangerous or offensive content. Use them at your own risk.

Credits and license

Model: Liquid AI.
MLX conversion: Axly's Customs, for the Erato app.

LFM Open License v1.0; see LICENSE. Commercial use is limited to organisations under the revenue threshold in the license.

Downloads last month: 91

Safetensors

Model size

3B params

Tensor type

U32

BF16

MLX

Hardware compatibility

4-bit

Model tree for Anotos/LFM2-2.6B-MLX-4bit

Base model

LiquidAI/LFM2-2.6B

Quantized

(22)

this model