1bit-MONSTER/DeepSeek-Coder-V2-Lite-Instruct-GGUF overview
DeepSeek Coder V2 Lite Instruct — GGUF Q4 K M bartowski's Q4 K M GGUF for DeepSeek's coding MoE ~16B total , re hosted with measured performance for the 1bit e…
Runs locally from ~9.65 GB disk (12 GB VRAM class GPUs with llama.cpp / guIDE).
Repository Files & Downloads
| File | Type | Quantization | Size | Link |
|---|---|---|---|---|
| DeepSeek-Coder-V2-Lite-Instruct-Q4_K_M.gguf | GGUF | Q4_K_M | 9.65 GB | Download |
Model Details
| Model ID | 1bit-MONSTER/DeepSeek-Coder-V2-Lite-Instruct-GGUF |
|---|---|
| Author | 1bit-MONSTER |
| Pipeline | — |
| License | other |
| Base model | deepseek-ai/DeepSeek-Coder-V2-Lite-Instruct |
| Last modified | 2026-09-26T15:57:27.000Z |
Model README
---
license: other
license_name: deepseek-license
license_link: LICENSE
base_model:
- deepseek-ai/DeepSeek-Coder-V2-Lite-Instruct
tags:
- gguf
- moe
- coder
---
DeepSeek-Coder-V2-Lite-Instruct — GGUF (Q4_K_M)
bartowski's Q4_K_M GGUF for DeepSeek's coding MoE (~16B total), re-hosted with measured
performance for the 1bit engine on Strix Halo.
Licensed under the DeepSeek License Agreement — open and permissive
including commercial use, with a small set of use-based restrictions (misuse
prevention), not a plain permissive license like Apache/MIT.
Contents
DeepSeek-Coder-V2-Lite-Instruct-Q4_K_M.gguf
Measured performance (Strix Halo, Vulkan)
pp512: 2002 tok/s · tg128: 99.5 tok/s
Running it
1bit serve -m DeepSeek-Coder-V2-Lite-Instruct-Q4_K_M.gguf --device vulkanAttribution
- Model: deepseek-ai/DeepSeek-Coder-V2-Lite-Instruct.
- Quantization: bartowski/DeepSeek-Coder-V2-Lite-Instruct-GGUF.
Run 1bit-MONSTER/DeepSeek-Coder-V2-Lite-Instruct-GGUF with guIDE
Download guIDE — the AI-native code editor with local LLM inference and 69 built-in tools.
Source: Hugging Face · Compare models