Qwen3.8-Flash-Next-CRACK-MLX

Instructions to use 0xSojalSec/Qwen3.8-Flash-Next-CRACK-MLX with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use 0xSojalSec/Qwen3.8-Flash-Next-CRACK-MLX 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("0xSojalSec/Qwen3.8-Flash-Next-CRACK-MLX")
config = load_config("0xSojalSec/Qwen3.8-Flash-Next-CRACK-MLX")

# 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 0xSojalSec/Qwen3.8-Flash-Next-CRACK-MLX with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "0xSojalSec/Qwen3.8-Flash-Next-CRACK-MLX"

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": "0xSojalSec/Qwen3.8-Flash-Next-CRACK-MLX"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent

How to use 0xSojalSec/Qwen3.8-Flash-Next-CRACK-MLX 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 "0xSojalSec/Qwen3.8-Flash-Next-CRACK-MLX"

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 0xSojalSec/Qwen3.8-Flash-Next-CRACK-MLX

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use 0xSojalSec/Qwen3.8-Flash-Next-CRACK-MLX with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "0xSojalSec/Qwen3.8-Flash-Next-CRACK-MLX"

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 "0xSojalSec/Qwen3.8-Flash-Next-CRACK-MLX" \
  --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-CRACK-MLX

File size: 4,525 Bytes

416fe72

---
language:
- en
- zh
license: other
license_name: qwen-community-1.0
license_link: LICENSE
library_name: mlx
pipeline_tag: image-text-to-text
base_model: JANGQ-AI/Qwen3.8-Flash-Next-JANG_2L
base_model_relation: quantized
thumbnail: dealign_mascot.png
tags:
- mlx
- apple-silicon
- abliterated
- uncensored
- crack
- jang
- jang-2l
- vision-language
- video
- reasoning
- thinking
- agent
- tool-use
- qwen4_exp
- moe
- ngram-embedding
- harmbench
- mmlu
- imatrix
- awq
---

<p align="center">
  <img src="./dealign_mascot.png" alt="dealignai" width="220">
</p>

# Qwen 3.8 Flash Next — CRACK-JANG2L

**CRACK abliterated** · **JANG_2L low-precision (MLX affine)** · Vision + Video language head · Reasoning (off / low / xhigh) · Agentic tools · Native MTP head · ~65 GB disk / **~47 GB active RAM** (the ~18 GB PLE hashed n-gram embedding table stays on disk and is streamed in on demand by vMLX's SSD-backed PLE runtime)

CRACK-abliterated build of [JANGQ-AI/Qwen3.8-Flash-Next-JANG_2L](https://huggingface.co/JANGQ-AI/Qwen3.8-Flash-Next-JANG_2L) — the smallest JANG tier of Alibaba's qwen4_exp preview (~176B mixture-of-experts, 512 experts, 6B active, Gated DeltaNet + Qwen Sparse Attention hybrid, hashed n-gram embedding, native multi-token-prediction head, vision + video). Refusal behavior is removed while deliberation, tool use, and multimodal capability are preserved.

> Research artifact. Download implies you accept responsibility for how the weights are used.

## Quality

Full benchmark tables (MMLU baseline vs CRACK vs Δ per subject + HarmBench-320 4-bucket per category × tier) will be added once the evaluation suite finishes. Preliminary partial results from the reasoning-off tier already show real-harm ASR **~99.4%** (317/319 TRUE_COMPLY, 0 hard-refuse) — significantly stronger compliance than the higher-precision siblings (CRACK-6S 92.5%, CRACK-JANG4M 91.6% at the same tier). MMLU delta pending.

Smoke tests confirmed: no code / math loop, coherent Fibonacci + product-rule derivative + integral at reasoning-off; full compliance with detailed methamphetamine synthesis routes at reasoning-off (no soft-refuse). The lower bit width appears to make the abliteration signal dominate the residual space more decisively.

## Multimodal + reasoning

- **Vision** — image comprehension intact.
- **Video** — video tower preserved from base.
- **Reasoning** — chat / think / max modes all intact. Control via `chat_template_kwargs: {"enable_thinking": true, "reasoning_effort": "low|high|xhigh"}`.
- **Tool calling** — Qwen XML parser (`tool_parser: "qwen"`). Tool call turns emit `<function=name><parameter=…>` inside `<tool_call>`.
- **Native MTP head** preserved and CRACK'd. Enable at serve time via `--native-mtp-depth N`.

## Runtime

Best experienced in **[vMLX](https://vmlx.net)** — the MLX inferencer with mixed-precision JANG, KV-cache quantization, prefix-cache reuse, agentic tool calling, and native MTP.

```
vmlx-engine serve dealignai/Qwen3.8-Flash-Next-CRACK-JANG2L --port 8888
```

Fits comfortably in ~64 GB of RAM (Apple Silicon), leaving room for KV cache and other workloads.

## Sampler

Vendor defaults:

```
temperature = 0.7    top_p = 0.9    top_k = 20
```

Greedy (temp=0) also works and is the mode CRACK compliance was measured at.

## Files

- `model-000{01..19}-of-00019.safetensors` — JANG low-precision shards
- `config.json`, `generation_config.json`, `chat_template.jinja` — vendor originals (unchanged)
- `tokenizer.json`, `tokenizer_config.json`, `merges.txt`, `vocab.json` — vendor tokenizer
- `SHARD_HASHES.txt` — SHA-256 of every shard for post-download verification
- `BENCHMARKS.json` — machine-readable eval scores (populated as evals complete)
- `LICENSE` — Qwen Community License 1.0

## Verify shards

```
cd /path/to/download
shasum -a 256 -c SHARD_HASHES.txt
```

All 19 shards should report `OK`.

## Related

- Base model: [JANGQ-AI/Qwen3.8-Flash-Next-JANG_2L](https://huggingface.co/JANGQ-AI/Qwen3.8-Flash-Next-JANG_2L) (unmodified quant reference)
- Siblings: [dealignai/Qwen3.8-Flash-Next-CRACK-6S](https://huggingface.co/dealignai/Qwen3.8-Flash-Next-CRACK-6S) (top JANG tier) · [dealignai/Qwen3.8-Flash-Next-CRACK-JANG4M](https://huggingface.co/dealignai/Qwen3.8-Flash-Next-CRACK-JANG4M) (mid tier)

---

[Ko-fi](https://ko-fi.com/dealignai) · [𝕏 @dealignai](https://x.com/dealignai) · [dealign.ai](https://dealign.ai)

<p align="center"><img src="./dealign_logo.png" alt="dealignai" width="140"></p>