DeepSeek V3
Open Source
About
DeepSeek V3: Latest flagship model. 685B total with MoE. 128K context. Open-source.
Capabilities
VisionMultimodalReasoningFunction CallingTool UseStructured OutputsCode Execution
Providers(11)
Compare all →| Provider | Input (per 1M) | Output (per 1M) | Type | |
|---|---|---|---|---|
| DeepInfra | $32 | $89 | Serverless | |
| Fireworks AI | $0.56 | $1.68 | Serverless | |
| DeepSeek Platform | $0.14 | $0.28 | Serverless | |
| Microsoft Foundry | — | — | ServerlessProvisioned | |
| OpenRouter | $0.26 | $0.38 | Serverless | |
| NVIDIA NIM | — | — | Serverless | |
| AWS Bedrock | — | — | Serverless | |
| Together AI | — | — | Serverless | |
| Bitdeer AI | $0.1 | $0.3 | Serverless | |
| SiliconFlow | $0.15 | $0.5 | Serverless | |
| Replicate API | $1.45 | $1.45 | Serverless |
Benchmark Scores(8)
| Benchmark | Score | Version | Source |
|---|---|---|---|
| HellaSwag | 95.7 | 10-shot | Open LLM Leaderboard, DeepSeek official |
| HumanEval | 85.5 | pass@1 | Open LLM Leaderboard, DeepSeek official |
| Massive Multitask Language Understanding | 88.5 | 5-shot | Open LLM Leaderboard |
| LiveCodeBench | 49.6 | 2026-04 | https://livecodebench.github.io/performances_generation.json |
| Aider Polyglot | 48.4 | 2026-04 | https://aider.chat/docs/leaderboards |
| BigCodeBench | 50.0 | 2025-01 (Instruct Pass@1) | https://bigcode-bench.github.io/results.json |
| Chatbot Arena | 1302.0 | — | https://lmarena.ai |
| MMLU PRO | 75.9 | — | https://huggingface.co/spaces/TIGER-Lab/MMLU-Pro |
Rankings
Specifications
FamilyDeepSeek V3
Released2024-12-26
Parameters671B
Context64k
ArchitectureMixture of Experts
Knowledge cutoff2024-04
Specializationgeneral
Trainingfinetuning
Created by
Advancing artificial general intelligence (AGI).
Hangzhou, Zhejiang, China
Founded 2023
Website