DeepSeek V3 0324
Specifications
- Input
- Output
- Context window
- 164K tokens
- Released
- Mar 2025
Performance
- Speed
- 27 t/s
- TTFT
- 3.4s
- Latency
- 1.4s
- Intelligence
- —
Pricing
- Input
- €0.30 per 1M tokens
- Output
- €0.95 per 1M tokens
About this model
Highly efficient DeepSeek flagship engineered for fast, capable reasoning and low-cost inference.
Technical specifications
- Capabilities
- Input modalities
- Output modalities
- Reasoning
- No
Knowledge horizon
Knowledge cutoff Jul 2024
Released Mar 2025
Today
Training to release 8 mo Since release 15 mo
See also
Add Model to Comparison
Search for a model to add