deepseek-ai
Text Generation
DeepSeek-V3
deepseek-ai/DeepSeek-V3
Tasks, architecture, licence and the memory this model needs before you pick hardware.
Understand this model
Review its task, architecture and access requirements. Missing specifications stay unknown.
Review model detailsEstimate hardware
Set precision, context and concurrency to explore the memory budget.
Calculate requirementsChoose where to run it
Explore cloud GPU offers, or discover models for your own computer.
Check your computerHosted API price
Use it through an API instead of renting a GPU
A hosted version is listed on OpenRouter. Prices are per million tokens and can change. Confirm on their page before relying on them.
- Input
- $0.32/ 1M tokens
- Output
- $0.89/ 1M tokens
- Context
- 163,840tokens
Model details
- Publisher
- deepseek-ai
- Family
- Unknown
- Task
- Text Generation
- Parameters
- 684,531,386,000
- Access
- Public
Licence
Licence information is unavailable.
Tasks and modalities
From the model config
Architecture
- Architecture names
- DeepseekV3ForCausalLM
- Experts selected per token
- 8
- Model type
- deepseek_v3
- Source-reported quantization method
- fp8
Ready to host it?
Estimate memory with your context length and concurrency, then see which GPU offers fit and what they cost.