DeepSeek-V3
Open SourceDeepSeek · 2025-05
DeepSeek's latest flagship model with 685B MoE architecture and exceptional coding performance.
Quick Facts
Parameters
685B total (37B active per token)
Context Window
128K tokens
Modalities
text
Open Source
Yes
License
MIT
Pricing
Free / API from $0.27/M tokens
Released
2025-05
Developer
DeepSeek
About
DeepSeek-V3 is DeepSeek's latest flagship large language model featuring 685 billion total parameters with Mixture-of-Experts architecture (37B active per token). It represents a significant advancement over DeepSeek-R1, with improved general knowledge, superior coding capabilities, and enhanced conversational abilities. DeepSeek-V3 demonstrates competitive performance against GPT-4o and Claude 3.5 Sonnet on key benchmarks while maintaining exceptional cost efficiency. Its open-weight availability under a permissive license continues DeepSeek's commitment to accessible AI.
Strengths
- +Open-weight with permissive MIT license
- +685B MoE architecture for strong performance
- +Exceptional coding and reasoning benchmarks
- +Highly cost-effective API pricing
Weaknesses
- −Text-only model, no vision or multimodal
- −Conversational polish still behind GPT-4o
- −Server availability can be inconsistent
Best For
Self-hosting powerful AI on own infrastructure
Complex coding and algorithm challenges
Cost-effective API integration at scale
Research and experimentation with open-weight models
Pricing
Free Chat
$0
- Unlimited DeepSeek chat
- V3 model
- File uploads
API
From $0.27/M tokens
- V3 API
- Rate limits
- Fine-tuning available
Self-Hosted
Free (open-weight)
- Full model weights
- Custom deployment
- Unlimited usage
Benchmarks
| Benchmark | DeepSeek-V3 | Competitor |
|---|---|---|
| MMLU | 88.5% | GPT-4o: 88.7% |
| HumanEval | 90.5% | Claude 3.5 Sonnet: 92.0% |
Technical Specs
Parameters
685B total (37B active per token)
Context Window
128K tokens
Modalities
text
Languages
Open Source
Yes
License
MIT
Related Models
GPT-4o
OpenAI
OpenAI's flagship multimodal model combining text, vision, and audio in one unified interface.
GPT-5
OpenAI
OpenAI's latest flagship model with enhanced reasoning, larger context, and improved multimodality.
Claude 3.5 Sonnet
Anthropic
Anthropic's balanced model offering strong reasoning, coding, and long-context capabilities.
Claude 4 Opus
Anthropic
Anthropic's most powerful model for complex reasoning, research, and specialized tasks.