Skip to content

DeepSeek V3

deepseek-ai/DeepSeek-V3
Open Source · chat · open-weights
Open Alert me on changes
Intelligence
Context
163.8K
Max output
Weights
Open
API $/1M
$0.26 / $1.03
Modalities
text
Released
25 Dec 2024
Intelligence Index via Artificial Analysis · 0–100, higher is better
Download image Share on X Share on LinkedIn
AI summary
● machine-written

DeepSeek V3: 685B mixture-of-experts open-weights model

DeepSeek V3 is a 685-billion parameter mixture-of-experts model released by DeepSeek in December 2024, available as open weights. Pre-trained on nearly 15 trillion tokens, it supports text generation with a 163,840 token context window. The model is available through multiple API providers with varying pricing structures.

What's new
  • Released December 25, 2024
  • 685B-parameter mixture-of-experts architecture
  • Pre-trained on nearly 15 trillion tokens
  • Supports tool calling and structured output features
  • Context window of 163,840 tokens
Best for
Instruction following and coding tasksChat and conversational applicationsTool-calling and agentic workflowsReasoning tasks
Sources

Source: https://huggingface.co/deepseek-ai/DeepSeek-V3