Instructions to use Jundot/Qwen3.8-Flash-Next-oQ3.5e-mtp with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Jundot/Qwen3.8-Flash-Next-oQ3.5e-mtp 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("Jundot/Qwen3.8-Flash-Next-oQ3.5e-mtp")
config = load_config("Jundot/Qwen3.8-Flash-Next-oQ3.5e-mtp")

# 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 Jundot/Qwen3.8-Flash-Next-oQ3.5e-mtp with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "Jundot/Qwen3.8-Flash-Next-oQ3.5e-mtp"

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": "Jundot/Qwen3.8-Flash-Next-oQ3.5e-mtp"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent

How to use Jundot/Qwen3.8-Flash-Next-oQ3.5e-mtp 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 "Jundot/Qwen3.8-Flash-Next-oQ3.5e-mtp"

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 Jundot/Qwen3.8-Flash-Next-oQ3.5e-mtp

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use Jundot/Qwen3.8-Flash-Next-oQ3.5e-mtp with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "Jundot/Qwen3.8-Flash-Next-oQ3.5e-mtp"

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 "Jundot/Qwen3.8-Flash-Next-oQ3.5e-mtp" \
  --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"

Qwen3.8-Flash-Next-oQ3.5e-mtp

Updated 2026-10-24: Re-quantized with Better Low-Bit oQ. Please re-download if you have an earlier copy.

This model was quantized using oQ (oMLX v0.7.3.dev1) mixed-precision quantization with Better Low-Bit oQ (jundot/omlx#4776).

A mixed-precision MLX checkpoint of Qwen/Qwen3.8-Flash-Next, an experimental preview of the architecture behind Qwen4: a mixture-of-experts vision-language model with 125B parameters and 6B activated, plus a 51B n-gram embedding and a 4B MTP module, that pairs Gated DeltaNet with Qwen Sparse Attention (QSA) and uses Gated Residual streams. It includes the language model, the n-gram embedding, the vision encoder and the multi-token prediction (MTP) head.

The effective average is 4.035 bits/weight excluding the N-gram embedding table, or 4.310 bits/weight for the complete checkpoint.

Quantization and Bit Distribution

oQ3.5e allocates bits by measured layer sensitivity, then rounds each group with an importance matrix (1024 samples x 512 tokens) and a weighted least-squares refit of its scale and bias. Below oQ4, the N-gram embedding table stays at 4 bits and is left out of the bit budget, because oMLX can serve it from SSD instead of memory.

Storage format	Logical weights	Share of total	Tensor storage	Effective bits/weight
Affine 3-bit, group 64	80.531B	44.739%	32.812 GiB	3.50
Affine 4-bit, group 64	42.782B	23.768%	22.412 GiB	4.50
Affine 5-bit, group 128	0.566B	0.315%	0.346 GiB	5.25
Affine 6-bit, group 64	1.608B	0.893%	1.217 GiB	6.50
Affine 8-bit, group 128	0.241B	0.134%	0.231 GiB	8.25
Affine 8-bit, group 64	2.556B	1.420%	2.529 GiB	8.50
I64	0.000B	0.000%	0.000 GiB	64.00
BF16	0.516B	0.287%	0.961 GiB	16.00
N-gram affine 4-bit, group 32	51.200B	28.445%	29.802 GiB	5.00
Total	180.000B	100%	90.311 GiB	4.310

Storage Breakdown

Component	Logical weights	Storage
Language backbone	125.744B	62.565 GB / 58.268 GiB
N-gram embedding	51.200B	32.000 GB / 29.802 GiB
Vision encoder	0.449B	0.898 GB / 0.836 GiB
MTP head	2.607B	1.508 GB / 1.404 GiB
Total	180.000B	96.971 GB / 90.311 GiB

License

Qwen Community License 1.0, following the source model. See LICENSE.

Downloads last month: 798

Safetensors

Model size

180B params

Tensor type

U32

BF16

I64

MLX

Hardware compatibility

3-bit

Model tree for Jundot/Qwen3.8-Flash-Next-oQ3.5e-mtp

Base model

Qwen/Qwen3.8-Flash-Next

Quantized

(524)

this model