DeepSeek · Code Models

DeepSeek-Coder-V2 Lite

Name: DeepSeek-Coder-V2 Lite
Author: DeepSeek

DeepSeek's open code model family; the Lite version uses a 16B mixture-of-experts with 2.4B active parameters and supports 338 programming languages.

Official page

Run it locally

With Ollama installed, download and start the default version with:

ollama run deepseek-coder-v2:16b

Variant	Download
deepseek-coder-v2:16b	8.9GB
deepseek-coder-v2:236b	133GB

Memory: About 8.9GB for the default deepseek-coder-v2:16b version (4-bit)
Formats: Safetensors (Hugging Face), GGUF (Ollama, llama.cpp, LM Studio)

The download size is roughly the memory (RAM or GPU VRAM) needed to load a variant; allow extra for long contexts. See our hardware guide.

Overview

DeepSeek-Coder-V2 was released in June 2024 in Lite (16B) and full (236B) versions, continuing pretraining of DeepSeek-V2 on code and math.

Highlights

16B total, 2.4B active parameters (Lite)
128K-token context window
338 programming languages
Fast inference for its quality

Typical uses

Local code completion and chat, code review and explanation.

Things to consider

DeepSeek model license applies.
Newer coder models have since appeared.
Download sizes refer to the default 4-bit quantized versions in the Ollama library; full-precision weights are larger.

Specifications are taken from the model card on Hugging Face, the Ollama library and the provider’s announcement. Benchmark results depend on evaluation settings; see how AI benchmarks work.

#Code Models #Open Weights #Local AI #DeepSeek