DeepSeek/Mixture of Experts

DeepSeek-Coder-V2 236B

Name: DeepSeek-Coder-V2 236B
Author: DeepSeek

Full-size MoE code model. 236B (21B active). 338 languages. 128K context.

codingreasoning

236B

Parameters (21B active)

160K

Context length

Benchmarks

Quantizations

Architecture

MoE

Released

2024-06-01

Layers

KV Heads

128

Head Dim

128

Family

deepseek

Quantization Options

Quant	Bits	VRAM	Quality
IQ2_XXS	2.38	70.7 GB	low
IQ2_M	2.93	86.9 GB	low
Q2_K	3.16	93.7 GB	low
IQ3_XXS	3.25	96.4 GB	low
IQ3_XS	3.5	103.7 GB	low
Q3_K_S	3.64	107.9 GB	low
IQ3_M	3.76	111.4 GB	low
Q3_K_M	4	118.5 GB	low
Q3_K_L	4.3	127.3 GB	moderate
IQ4_XS	4.46	132.1 GB	moderate
Q4_K_S	4.67	138.3 GB	moderate
Q4_K_M	4.89	144.7 GB	good
Q5_K_S	5.57	164.8 GB	good
Q5_K_M	5.7	168.6 GB	good
Q6_K	6.56	194.0 GB	excellent
Q8_0	8.5	251.2 GB	lossless
FP16	16	472.5 GB	lossless

Select your GPU above to see speed estimates and compatibility for each quantization.

▸ READY TO RUN THIS?RENT BY THE HOUR

RENT A GPU AND RUN DEEPSEEK-CODER-V2 236B NOW

Rent on RunPod →Or Vast.ai →

Spin up an A100 / H100 / 4090 in ~60s. Pay by the second. Cancel anytime.

Community Ratings

Loading ratings...

Benchmarks (8)

HumanEval90.2

BBH83.9

MBPP76.2

MATH75.7

MATH-50074.3

LiveCodeBench43.4

BigCodeBench40.4

AA Intelligence10.6

Run this model

▸Easiest way to get started·Beginners

DOCS ↗

curl -fsSL https://ollama.com/install.sh | sh

$ollama run deepseek:236b-q4_K_M

Tag may need adjustment — check ollama.com/library/deepseek for available tags.

▸ SETUP GUIDE

Auto-setup with fitmyllm CLI

Detects your GPU, recommends the best model, downloads it, and starts chatting — zero config. Benchmarks your speed and contributes anonymous data to improve predictions.

pip install fitmyllmthen run fitmyllmLearn more

Auto-detect GPULive tok/s in chatSpeed benchmarks9 inference engines

HuggingFace GGUF Downloads Build Hardware