DeepSeek V3 (Dec '24)
DeepSeekDeepSeekOpen WeightMIT + Model License (Commercial use allowed)
Description
A powerful Mixture-of-Experts (MoE) language model with 671B total parameters (37B activated per token). Features Multi-head Latent Attention (MLA), auxiliary-loss-free load balancing, and multi-token prediction training. Pre-trained on 14.8T tokens with strong performance in reasoning, math, and code tasks.
Release Date
2024-12-26
Parameters
671.0B
Context Length
—
Modalities
text
Capability Radar
70
general
50
coding
70
reasoning
51
scienceest.
64
agents
0
multimodal
Science is estimated from LLM Stats science scores or reasoning when no dedicated science benchmarks are available.
Rankings
No ranking data available
Benchmark Scores (LLM Stats)
(LLM Stats (zeroeval))Chat
IFEvalGoogle Research (2023)
86.1%SR
Factuality
SimpleQA
24.9%SR
General
MMLU
88.5%SR
C-Eval
86.5%SR
Aider-Polyglot Edit
79.7%SR
CSimpleQA
64.8%SR
Aider-Polyglot
49.6%SR
Language
CLUEWSC
90.9%SR
MMLU-Redux
89.1%SR
MMLU-Pro
75.9%SR
Long Context
LongBench v2
48.7%SR
Math
MATH-500
90.2%SR
CNMO 2024
43.2%SR
AIME 2024
39.2%SR
Reasoning
DROP
91.6%SR
HumanEval-Mul
82.6%SR
FRAMES
73.3%SR
GPQANYU + Cohere + Anthropic (2023)
59.1%SR
SWE-Bench Verified
42.0%SR
LiveCodeBench
37.6%SR
AA Evaluation Indices
(Artificial Analysis)No AA evaluation data available
LLM Stats Category Scores
(LLM Stats (zeroeval))Chat90
Instruction Following90
Language80
Legal80
Finance80
Healthcare80
Math70
Reasoning70
Search70
Structured Output70
General70
Physics60
Biology60
Chemistry60
Long Context50
Code50
Frontend Development40
Factuality20
Pricing
Input Price$0.32 / 1M tokens
Output Price$0.89 / 1M tokens
Blended Price (3:1)$0.4625 / 1M tokens
Speed
Tokens/sec73.2
Time to First Token0.87s
Time to Answer0.87s
Provider Price Ranking
Provider Price Ranking
5 providers
Cheapest: Alibaba (China)Most Expensive: Helicone
ProviderInputOutput
1Alibaba (China)Cheapest
$0.287
$1.147
2Qiniu
$0.28986
$1.15942
3DeepSeekPRIMARY
$0.32
$0.89
4DeepInfra
$0.32
$0.89
5Helicone
$0.56
$1.68
Compare pricing across different API providers for this model.