FitMyLLM
DeepSeek/Mixture of Experts

DeepSeekDeepSeek V2 Lite 16B

DeepSeek V2 Lite — efficient MoE for chat and coding at low cost.

chatcoding
16B
Parameters (2.4B active)
31K
Context length
7
Benchmarks
10
Quantizations
0
Architecture
MoE
Released
2024-05-06
Layers
27
KV Heads
16
Head Dim
128
Family
deepseek

Quantization Options

Context length:
QuantBitsVRAM @ 16KQuality
Q3_K_M4
8.8 GB
8.5 + 0.4 KV
low
Q3_K_L4.3
9.4 GB
9.1 + 0.4 KV
moderate
IQ4_XS4.46
9.8 GB
9.4 + 0.4 KV
moderate
Q4_K_S4.67
10.2 GB
9.8 + 0.4 KV
moderate
Q4_K_M4.89
10.6 GB
10.3 + 0.4 KV
good
Q5_K_S5.57
12.0 GB
11.6 + 0.4 KV
good
Q5_K_M5.7
12.2 GB
11.9 + 0.4 KV
good
Q6_K6.56
14.0 GB
13.6 + 0.4 KV
excellent
Q8_08.5
17.8 GB
17.5 + 0.4 KV
lossless
FP1616
32.8 GB
32.5 + 0.4 KV
lossless

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

Deploying for a team or in production? Size GPUs, cost & scaling in Enterprise →
▸ READY TO RUN THIS?RENT BY THE HOUR

RENT A GPU AND RUN DEEPSEEK V2 LITE 16B NOW

Rent on Vast.ai →Or RunPod →

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

Community Ratings

Loading ratings...

Benchmarks (7)

MATH57.0
IFEval43.8
MMLU-PRO40.7
BBH40.7
BigCodeBench36.8
MUSR28.7
GPQA18.3

Run this model

▸Easiest way to get started·Beginners
DOCS ↗
curl -fsSL https://ollama.com/install.sh | sh
$ollama run deepseek-v2:16b-lite-chat-q4_K_M

Downloads and runs automatically. Add --verbose for speed stats.

▸ 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
HuggingFaceOllama LibraryGGUF DownloadsBuild Hardware

GPUs that can run this model

At Q4_K_M quantization. Sorted by minimum VRAM.

NVIDIA RTX 2080 Ti
11 GB VRAM • 616 GB/s
NVIDIA
$350
AmazonNeweggeBayPCPartPickerRent on RunPodRent on Vast.ai
NVIDIA GTX 1080 Ti
11 GB VRAM • 484 GB/s
NVIDIA
$200
AmazonNeweggeBayPCPartPickerRent on RunPodRent on Vast.ai
NVIDIA GeForce GTX 1080 Ti
11 GB VRAM • 484 GB/s
NVIDIA
AmazonNeweggeBayPCPartPickerRent on RunPodRent on Vast.ai
NVIDIA GeForce RTX 2080 Ti
11 GB VRAM • 616 GB/s
NVIDIA
$225
AmazonNeweggeBayPCPartPickerRent on RunPodRent on Vast.ai
NVIDIA RTX 5070
12 GB VRAM • 672 GB/s
NVIDIA
$549
AmazonNeweggeBayPCPartPickerRent on RunPodRent on Vast.ai
NVIDIA RTX 4070 Ti
12 GB VRAM • 504 GB/s
NVIDIA
$799
AmazonNeweggeBayPCPartPickerRent on RunPodRent on Vast.ai
NVIDIA RTX 4070 SUPER
12 GB VRAM • 504 GB/s
NVIDIA
$599
AmazonNeweggeBayPCPartPickerRent on RunPodRent on Vast.ai
NVIDIA RTX 4070
12 GB VRAM • 504 GB/s
NVIDIA
$549
AmazonNeweggeBayPCPartPickerRent on RunPodRent on Vast.ai
NVIDIA RTX 3080 Ti
12 GB VRAM • 912 GB/s
NVIDIA
$550
AmazonNeweggeBayPCPartPickerRent on RunPodRent on Vast.ai
NVIDIA RTX 3080 12GB
12 GB VRAM • 912 GB/s
NVIDIA
$599
AmazonNeweggeBayPCPartPickerRent on RunPodRent on Vast.ai
NVIDIA RTX 3060 12GB
12 GB VRAM • 360 GB/s
NVIDIA
$329
AmazonNeweggeBayPCPartPickerRent on RunPodRent on Vast.ai
AMD RX 7700 XT
12 GB VRAM • 432 GB/s
AMD
$449
AmazonNeweggeBayPCPartPickerRent on RunPodRent on Vast.ai
AMD RX 6700 XT
12 GB VRAM • 384 GB/s
AMD
$344
AmazonNeweggeBayPCPartPickerRent on RunPodRent on Vast.ai
AMD RX 6750 XT
12 GB VRAM • 432 GB/s
AMD
$299
AmazonNeweggeBayPCPartPickerRent on RunPodRent on Vast.ai
Intel Arc B580
12 GB VRAM • 456 GB/s
INTEL
$249
AmazonNeweggeBayPCPartPickerRent on RunPodRent on Vast.ai
NVIDIA Tesla K40c
12 GB VRAM • 288 GB/s
NVIDIA
NVIDIA Tesla K40d
12 GB VRAM • 288 GB/s
NVIDIA
NVIDIA Tesla K40m
12 GB VRAM • 288 GB/s
NVIDIA
NVIDIA Tesla K40s
12 GB VRAM • 288 GB/s
NVIDIA
NVIDIA Tesla K40st
12 GB VRAM • 288 GB/s
NVIDIA
NVIDIA Tesla K40t
12 GB VRAM • 288 GB/s
NVIDIA
NVIDIA Tesla K80
12 GB VRAM • 241 GB/s
NVIDIA
NVIDIA Tesla M40
12 GB VRAM • 288 GB/s
NVIDIA
NVIDIA Tesla P100 PCIe 12 GB
12 GB VRAM • 549 GB/s
NVIDIA
NVIDIA GeForce RTX 2060 12 GB
12 GB VRAM • 336 GB/s
NVIDIA
$140
AmazonNeweggeBayPCPartPickerRent on RunPodRent on Vast.ai
NVIDIA GeForce RTX 3060 12 GB
12 GB VRAM • 360 GB/s
NVIDIA
$329
AmazonNeweggeBayPCPartPickerRent on RunPodRent on Vast.ai
NVIDIA
$329
AmazonNeweggeBayPCPartPickerRent on RunPodRent on Vast.ai
NVIDIA GeForce RTX 3080 Ti
12 GB VRAM • 912 GB/s
NVIDIA
$1199
AmazonNeweggeBayPCPartPickerRent on RunPodRent on Vast.ai
NVIDIA RTX A2000 12 GB
12 GB VRAM • 288 GB/s
NVIDIA
$550
AMD Radeon RX 6700 XT
12 GB VRAM • 384 GB/s
AMD
$250
AmazonNeweggeBayPCPartPickerRent on RunPodRent on Vast.ai

Find the best GPU for DeepSeek V2 Lite 16B

Build Hardware for DeepSeek V2 Lite 16B

Model Card

View on HuggingFace

DeepSeek V2 Lite — efficient MoE for chat and coding at low cost.

▸ SPEC SHEET

DeepSeek V2 Lite 16B — 16B MoE.

▸ SPECIFICATIONS
PARAMETERS
16B (2.4B active)
ARCHITECTURE
Mixture of Experts
CONTEXT LENGTH
31K tokens
CAPABILITIES
chat, coding
RELEASE DATE
2024-05-06
PROVIDER
DeepSeek
FAMILY
deepseek
▸ VRAM REQUIREMENTS
QUANTBPWVRAMQUALITY
Q3_K_M48.5 GB88%
Q3_K_L4.39.1 GB90%
IQ4_XS4.469.4 GB92%
Q4_K_S4.679.8 GB93%
Q4_K_M4.8910.3 GB94%
Q5_K_S5.5711.6 GB96%
Q5_K_M5.711.9 GB96%
Q6_K6.5613.6 GB97%
Q8_08.517.5 GB100%
FP161632.5 GB100%
§ 01BENCHMARK SCORES
MMLU-PRO40.7
MATH57.0
IFEval43.8
BBH40.7
GPQA18.3
MUSR28.7
BigCodeBench36.8
§ 02RUN COMMAND

Run DeepSeek V2 Lite 16B locally with Ollama — needs 10.3 GB VRAM at Q4_K_M:

$ollama run deepseek-v2:16b
§ 03COMPATIBLE GPUs
30 @ Q4_K_M
NVIDIA RTX 2080 Ti
11 GB · 616 GB/s
NVIDIA GTX 1080 Ti
11 GB · 484 GB/s
NVIDIA GeForce GTX 1080 Ti
11 GB · 484 GB/s
NVIDIA GeForce RTX 2080 Ti
11 GB · 616 GB/s
NVIDIA RTX 5070
12 GB · 672 GB/s
NVIDIA RTX 4070 Ti
12 GB · 504 GB/s
NVIDIA RTX 4070 SUPER
12 GB · 504 GB/s
NVIDIA RTX 4070
12 GB · 504 GB/s
NVIDIA RTX 3080 Ti
12 GB · 912 GB/s
NVIDIA RTX 3080 12GB
12 GB · 912 GB/s
NVIDIA RTX 3060 12GB
12 GB · 360 GB/s
AMD RX 7700 XT
12 GB · 432 GB/s
AMD RX 6700 XT
12 GB · 384 GB/s
AMD RX 6750 XT
12 GB · 432 GB/s
Intel Arc B580
12 GB · 456 GB/s