Instructions to use ToPo-ToPo/Ornith-1.5-122B-mlx-8bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use ToPo-ToPo/Ornith-1.5-122B-mlx-8bit 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("ToPo-ToPo/Ornith-1.5-122B-mlx-8bit")
config = load_config("ToPo-ToPo/Ornith-1.5-122B-mlx-8bit")

# 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 ToPo-ToPo/Ornith-1.5-122B-mlx-8bit with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "ToPo-ToPo/Ornith-1.5-122B-mlx-8bit"

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": "ToPo-ToPo/Ornith-1.5-122B-mlx-8bit"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent

How to use ToPo-ToPo/Ornith-1.5-122B-mlx-8bit 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 "ToPo-ToPo/Ornith-1.5-122B-mlx-8bit"

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 ToPo-ToPo/Ornith-1.5-122B-mlx-8bit

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use ToPo-ToPo/Ornith-1.5-122B-mlx-8bit with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "ToPo-ToPo/Ornith-1.5-122B-mlx-8bit"

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 "ToPo-ToPo/Ornith-1.5-122B-mlx-8bit" \
  --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"

ToPo-ToPo/Ornith-1.5-122B-mlx-8bit

MLX 8bit conversion of ornith-ai/Ornith-1.5-122B for Apple Silicon (mlx-vlm), with the vision tower kept.

Provenance (self-converted from official weights)

Source: ornith-ai/Ornith-1.5-122B (bf16, license: mit)
Tool: mlx-vlm 0.7.6 (mlx 0.32.3) — equivalent to mlx_vlm.convert --hf-path ornith-ai/Ornith-1.5-122B --mlx-path . -q --q-bits 8 --q-group-size 64
Quantization: affine, 8 bits, group size 64 (router gates included)
Effective: 8.36 bits/weight (130.6 GB)
Vision: the source checkpoint is Qwen3_5MoeForConditionalGeneration and contains the vision tower (333 tensors) and the image/video processor configs. This conversion keeps them, so the model accepts image input.
The MTP (multi-token prediction) weights of the source are not included here. They are published separately as a drafter: ToPo-ToPo/Ornith-1.5-122B-MTP-bf16.
Validation: served through local-llm-server (mlx-vlm 0.7.9 backend) with the MTP drafter on an Apple M3 Ultra (256 GB). Japanese chat and multi-step tool calling worked; decoding ran at about 29–44 tokens/s (about 130 GB resident).

Usage

from mlx_vlm import load, generate
model, processor = load("ToPo-ToPo/Ornith-1.5-122B-mlx-8bit")

With the MTP drafter:

mlx_vlm.generate --model ToPo-ToPo/Ornith-1.5-122B-mlx-8bit \
  --draft-model ToPo-ToPo/Ornith-1.5-122B-MTP-bf16 --draft-kind mtp \
  --prompt "..." --max-tokens 400

Downloads last month: -

Safetensors

Model size

123B params

Tensor type

U32

BF16

MLX

Hardware compatibility

8-bit

Model tree for ToPo-ToPo/Ornith-1.5-122B-mlx-8bit

Base model

ornith-ai/Ornith-1.5-122B

Quantized

(44)

this model