Instructions to use mlx-community/Inkling-2-Small-oQ4e with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use mlx-community/Inkling-2-Small-oQ4e with MLX:

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

from mlx_vlm import load, generate
from mlx_vlm.prompt_utils import apply_chat_template
from mlx_vlm.utils import load_config

# Load the model
model, processor = load("mlx-community/Inkling-2-Small-oQ4e")
config = load_config("mlx-community/Inkling-2-Small-oQ4e")

# Prepare input
image = ["http://images.cocodataset.org/val2017/000000039769.jpg"]
prompt = "Describe this image."

# Apply chat template
formatted_prompt = apply_chat_template(
    processor, config, prompt, num_images=1
)

# Generate output
output = generate(model, processor, formatted_prompt, image)
print(output)

Notebooks
Google Colab
Kaggle
Local Apps Settings
LM Studio

How to use mlx-community/Inkling-2-Small-oQ4e with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "mlx-community/Inkling-2-Small-oQ4e"

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": "mlx-community/Inkling-2-Small-oQ4e"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent

How to use mlx-community/Inkling-2-Small-oQ4e 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 "mlx-community/Inkling-2-Small-oQ4e"

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 mlx-community/Inkling-2-Small-oQ4e

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use mlx-community/Inkling-2-Small-oQ4e with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "mlx-community/Inkling-2-Small-oQ4e"

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 "mlx-community/Inkling-2-Small-oQ4e" \
  --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"

Inkling-2-Small-oQ4e

This model was quantized using oQ (oMLX v0.8.0) mixed-precision quantization.

A mixed-precision MLX checkpoint of thinkingmachines/Inkling-2-Small, a multimodal mixture-of-experts model with 594B total and 17B active parameters. It takes text and image inputs with a context window of up to 1M tokens. It includes the language model, the vision encoder and the bundled DSpark draft in mtp.safetensors, which oMLX Lightning MTP uses for speculative decoding.

The effective average is 4.553 bits/weight for the language model and vision encoder, or 4.585 bits/weight including the DSpark draft.

Quantization and Bit Distribution

oQ4e keeps the routed experts of layers 0 to 39 in their source NVFP4 format and requantizes the MXFP8 experts of layers 40 to 47 to affine 4-bit. It stores the attention projections, the embeddings and the vocabulary head as affine 8-bit. The vision encoder and the DSpark draft keep BF16, and the router keeps FP32.

Storage format	Logical weights	Share of total	Tensor storage	Effective bits/weight
Affine 4-bit, group 64	97.845B	16.471%	51.258 GiB	4.50
NVFP4, group 16	489.224B	82.354%	256.289 GiB	4.50
Affine 8-bit, group 128	4.713B	0.793%	4.526 GiB	8.25
Affine 8-bit, group 64	1.853B	0.312%	1.833 GiB	8.50
BF16	0.331B	0.056%	0.617 GiB	16.00
FP32	0.085B	0.014%	0.319 GiB	32.00
Total	594.051B	100%	314.842 GiB	4.553

Storage Breakdown

Component	Logical weights	Storage
Language model	593.723B	337.404 GB / 314.232 GiB
Vision	0.328B	0.655 GB / 0.610 GiB
DSpark draft (MTP)	1.712B	3.423 GB / 3.188 GiB
Total	595.762B	341.483 GB / 318.031 GiB

License

Apache 2.0, following the source model.

Downloads last month: -

Safetensors

Model size

594B params

Tensor type

U32

BF16

F32

MLX

Hardware compatibility

4-bit

Model tree for mlx-community/Inkling-2-Small-oQ4e

Base model

thinkingmachines/Inkling-2-Small

Quantized

(5)

this model