All models
DeepSeek

DeepSeek V3

frontierOpen Source

DeepSeek

Chinese lab DeepSeek's impressive open-source flagship. Competitive with top-tier models at a dramatically lower inference cost.

Official page
open-sourcecost-effectivecoding
Context
163.84K
Input / 1M
$0.20
Output / 1M
$0.77
Knowledge cutoff
2024-07

Capabilities

Code5/5
Reasoning4/5
Multilingual3/5
Safety2/5
Speed4/5

Modalities & Features

Vision (image input) No
Audio No
Video No
Function / tool calling Yes
JSON mode Yes
Extended thinking No

Context & Output

Context window163.84K
Max output tokens8.192K
Knowledge cutoff2024-07
Release date2024-12

API & Access

Open source Yes
Streaming Yes
Batch API No
Prompt caching Yes
Fine-tuning No

Benchmarks

MMLUGeneral knowledge
88.5%
GPQA DiamondHard science Q&A
59.1%
MATHMath problem solving
75.9%
AIME 2025Advanced math
N/A
GSM8KGrade-school math
92.8%
HumanEvalCode generation
82.6%
LiveCodeBenchLive coding problems
N/A
SWE-bench VerifiedReal-world software
N/A
HellaSwagCommonsense inference
N/A

Source: provider technical reports & independent evaluations. N/A = not yet published for this model.

Pricing (per 1M tokens)

Input$0.20
Output$0.77
Consumer plan
DeepSeek Chat
Free
Prices shown are list prices and may vary by provider plan. Always check official documentation for the latest pricing.