DeepSeek V3 Specs, Benchmarks & Downloads
DeepSeek V3
DeepSeek foundation MoE model
Status
active
Parameters
671.00
Context Window
131072
Modality
Text
Release Date
Dec 26, 2024
About
Explore DeepSeek V3, a text AI model developed by DeepSeek. View specifications, context window, architecture, license, supported platforms, features, benchmarks, official downloads, and related articles.
DeepSeek V3 is a large mixture-of-experts model for reasoning and coding.
Family
DeepSeekDeveloper
DeepSeekArchitecture
Mixture of ExpertsLicense
DeepSeek LicenseHardware Requirements
Minimum:
Use Case: Distributed inference
GPU Minimum: 2x RTX 4090
GPU VRAM: 48GB
System RAM: 128GB
Storage: 200GB
Operating System: Linux
DeepSeek V3 large model
Recommended:
Use Case: Server inference
GPU Minimum: 4x A100
GPU VRAM: 320GB
System RAM: 512GB
Storage: 500GB
Operating System: Linux
Large scale deployment
High Performance:
Use Case: Enterprise serving
GPU Minimum: 8x H100
GPU VRAM: 640GB
System RAM: 1024GB
Storage: 1TB
Operating System: Linux
Maximum performance
Model Deployment Readiness
Model Ecosystem Support
12%
Platforms
17%
Features
0%
Downloads
20%
Platform Compatibility
17%
Supported across 2 of 12 platforms
Context Window
Industry-standard context window comparison
131,072
tokens
128K
4K
32K
128K
1M
2M
Database Rank
#1
Percentile
Top 100%
Capacity
Very High
Benchmark Performance
Evaluation profile across publicly reported benchmarks
|
HumanEval
|
Coding | 65.2 % | 2024 |
|
LiveCodeBench
|
Coding | 40.5 % | 2024 |
|
SWE-Bench
|
Software Engineering | 42.0 % | 2024 |
Highest Benchmark
HumanEval
65.2%
Benchmarks Available
3
Latest Evaluation
Dec 2024
