---
title: JanusCoder-8B-DeepSeek-Speciale-qx86-hi-mlx
canonical_url: "https://www.modelscope.cn/models/nightmedia/JanusCoder-8B-DeepSeek-Speciale-qx86-hi-mlx"
md_url: "https://www.modelscope.cn/models/nightmedia/JanusCoder-8B-DeepSeek-Speciale-qx86-hi-mlx.md"
repository: nightmedia/JanusCoder-8B-DeepSeek-Speciale-qx86-hi-mlx
last_updated: 2026-01-31
license: apache-2.0
pipeline_tag: text-generation
tasks:
  - text-generation
model_type:
  - qwen3
architectures:
  - Qwen3ForCausalLM
base_model:
  - unsloth/JanusCoder-8B
  - TeichAI/Qwen3-8B-DeepSeek-v3.2-Speciale-Distill
  - nightmedia/Qwen3-8B-Element6
base_model_relation: merge
parameters: 678.2M
tensor_type:
  - BF16
  - U32
library_name:
  - mlx
  - safetensors
  - pytorch
frameworks:
  - pytorch
language:
  - en
  - zh
inference_backends:
  - "deploy_task text/emb"
  - "lmdeploy_turbomind 0.9.1"
  - "sglang 0.5.2"
  - "vllm 0.9.2"
downloads: 54
stars: 0
tags:
  - coding
  - research
  - "deep thinking"
  - "128k context"
  - Qwen3
  - "All use cases"
  - creative
  - "creative writing"
  - "fiction writing"
  - "plot generation"
  - "sub-plot generation"
  - "story generation"
  - "scene continue"
  - storytelling
  - "fiction story"
  - "science fiction"
  - "all genres"
  - story
  - writing
  - "vivid prosing"
  - "vivid writing"
  - fiction
  - roleplaying
  - bfloat16
  - finetune
  - mergekit
  - merge
  - mlx
---

# JanusCoder-8B-DeepSeek-Speciale-qx86-hi-mlx

> JanusCoder-8B-DeepSeek-Speciale-qx86-hi-mlx - nightmedia 在 ModelScope 开源的模型。JanusCoder-8B-DeepSeek-Speciale-qx86-hi-mlx

nightmedia/JanusCoder-8B-DeepSeek-Speciale-qx86-hi-mlx 是 ModelScope 魔搭社区上的 678.2M 参数text-generation模型，采用 apache-2.0 许可，基于 unsloth/JanusCoder-8B、TeichAI/Qwen3-8B-DeepSeek-v3.2-Speciale-Distill、nightmedia/Qwen3-8B-Element6 构建，可用 deploy_task text/emb、lmdeploy_turbomind 0.9.1、sglang 0.5.2 部署。

- **Repository**: nightmedia/JanusCoder-8B-DeepSeek-Speciale-qx86-hi-mlx
- **License**: apache-2.0
- **Tasks**: text-generation
- **Parameters**: 678.2M
- **Base model**: unsloth/JanusCoder-8B, TeichAI/Qwen3-8B-DeepSeek-v3.2-Speciale-Distill, nightmedia/Qwen3-8B-Element6
- **Inference backends**: deploy_task text/emb, lmdeploy_turbomind 0.9.1, sglang 0.5.2, vllm 0.9.2
- **Tags**: coding, research, deep thinking, 128k context, Qwen3, All use cases, creative, creative writing, fiction writing, plot generation, sub-plot generation, story generation, scene continue, storytelling, fiction story, science fiction, all genres, story, writing, vivid prosing, vivid writing, fiction, roleplaying, bfloat16, finetune, mergekit, merge, mlx
- **Downloads**: 54
- **Stars**: 0
- **Last updated**: 2026-01-31

Source: https://www.modelscope.cn/models/nightmedia/JanusCoder-8B-DeepSeek-Speciale-qx86-hi-mlx

---

# JanusCoder-8B-DeepSeek-Speciale-qx86-hi-mlx

## Qwen3-8B-Element6

This model is a merge of
- unsloth/JanusCoder-8B
- TeichAI/Qwen3-8B-DeepSeek-v3.2-Speciale-Distill

Brainwaves
```brainwave
          arc   arc/e boolq hswag obkqa piqa  wino
qx86-hi   0.530,0.724,0.846,0.712,0.422,0.786,0.661

Janus     0.537,0.731,0.862,0.697,0.446,0.782,0.667
Element6  0.530,0.724,0.846,0.712,0.422,0.786,0.661

Perplexity
qx86-hi   4.361 ± 0.032
qx64-hi   4.492 ± 0.033
```

-G


## Use with mlx

```bash
pip install mlx-lm
```

```python
from mlx_lm import load, generate

model, tokenizer = load("JanusCoder-8B-DeepSeek-Speciale-qx86-hi-mlx")

prompt = "hello"

if tokenizer.chat_template is not None:
    messages = [{"role": "user", "content": prompt}]
    prompt = tokenizer.apply_chat_template(
        messages, add_generation_prompt=True, return_dict=False,
    )

response = generate(model, tokenizer, prompt=prompt, verbose=True)
```