---
title: DeepSeek-V4-Flash-MTP-DSpark-MLX
canonical_url: "https://www.modelscope.ai/models/inferencerlabs/DeepSeek-V4-Flash-MTP-DSpark-MLX"
md_url: "https://www.modelscope.ai/models/inferencerlabs/DeepSeek-V4-Flash-MTP-DSpark-MLX.md"
repository: inferencerlabs/DeepSeek-V4-Flash-MTP-DSpark-MLX
last_updated: 2026-11-13
pipeline_tag: text-generation
tasks:
  - text-generation
model_type:
  - deepseek_v4_dspark
base_model:
  - deepseek-ai/DeepSeek-V4-Flash-0731
base_model_relation: quantized
parameters: 4.3B
tensor_type:
  - BF16
  - U8
  - F32
  - U32
library_name:
  - mlx
  - safetensors
  - pytorch
frameworks:
  - pytorch
language:
  - en
downloads: 417
stars: 0
tags:
  - mlx
---

# DeepSeek-V4-Flash-MTP-DSpark-MLX

> DeepSeek-V4-Flash-MTP-DSpark-MLX - An open-source model by inferencerlabs on ModelScope. DeepSeek-V4-Flash-0731 DSpark See DeepSeek-V4-Flash with MTP in action: demonstration videos

inferencerlabs/DeepSeek-V4-Flash-MTP-DSpark-MLX is a 4.3B-parameter text-generation model on ModelScope. derived from deepseek-ai/DeepSeek-V4-Flash-0731.

- **Repository**: inferencerlabs/DeepSeek-V4-Flash-MTP-DSpark-MLX
- **Tasks**: text-generation
- **Parameters**: 4.3B
- **Base model**: deepseek-ai/DeepSeek-V4-Flash-0731
- **Tags**: mlx
- **Downloads**: 417
- **Stars**: 0
- **Last updated**: 2026-11-13

Source: https://www.modelscope.ai/models/inferencerlabs/DeepSeek-V4-Flash-MTP-DSpark-MLX

---

# DeepSeek-V4-Flash-0731 DSpark
See DeepSeek-V4-Flash with MTP in action: [demonstration videos](https://youtube.com/xcreate)

This draft model contains the extracted **DSpark** layers to be used alongside the [DeepSeek-V4-Flash-0731-MLX](https://huggingface.co/models?search=inferencerlabs/deepseek-v4-flash-0731-mlx) model as a speculative decoder for improved performance.

#### Tested on an M3 Ultra 512 GiB RAM using [Inferencer app v2.3.1](https://inferencer.com)
### Tetris HTML
<table style="border-collapse: collapse; border: none; text-align:left; margin-top:10px; margin-bottom:0px;">
  <tr><td><strong>Text inference</strong></td><td>~31 tokens/s @ 500 tokens ~148.5 GiB</td></tr>
  <tr><td><strong>Text inference (with MTP)</strong></td><td>~35.8 tokens/s @ 500 tokens ~152.9 GiB</td></tr>
  <tr><td><strong>Text inference (with DSpark)</strong></td><td>~53.1 tokens/s @ 500 tokens ~165.6 GiB</td></tr>
</table>

### Tell me about Paris
<table style="border-collapse: collapse; border: none; text-align:left; margin-top:10px; margin-bottom:0px;">
  <tr><td><strong>Text inference</strong></td><td>~30.5 tokens/s @ 500 tokens ~148.5 GiB</td></tr>
  <tr><td><strong>Text inference (with MTP)</strong></td><td>~37.3 tokens/s @ 500 tokens ~152.9 GiB</td></tr>
  <tr><td><strong>Text inference (with DSpark)</strong></td><td>~31.3 tokens/s @ 500 tokens ~165.6 GiB</td></tr>
</table>

<p style="margin-bottom:0;padding-bottom:0;">Enable speculative decoding in Inference Contols:</p>
<div style="margin-left:10px;">
  <img 
    src="https://cdn-uploads.huggingface.co/production/uploads/688479d616f1ec82fa645019/weJgx7wZI7VgOdQe827yS.jpeg"
    style="display:block;margin:0;">
</div>

##### Extracted with a modified version of [MLX-VLM](https://github.com/Blaizzy/mlx-vlm)
##### For more details see our [demonstration videos](https://youtube.com/xcreate) or visit [DeepSeek-V4-Flash](https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash).

## Disclaimer

We are not the creator, originator, or owner of any model listed. Each model is created and provided by third parties. Models may not always be accurate or contextually appropriate. You are responsible for verifying the information before making important decisions. We are not liable for any damages, losses, or issues arising from its use, including data loss or inaccuracies in AI-generated content.