Head to head

Llama 3.3 70B Versatile vs DeepSeek V4 Pro

On a blended workload of one million input tokens and three million output tokens, Llama 3.3 70B Versatile is 1.0× cheaper than DeepSeek V4 Pro.

Llama 3.3 70B VersatileDeepSeek V4 Pro
Provider Groq DeepSeek
Input / MTok $0.59 $0.435
Output / MTok $0.79 $0.87
Cache read / MTok $0.003625
Context window 131,072 1,000,000
Blended 1M in + 3M out $2.96 $3.04
Status ga ga

Llama 3.3 70B Versatile detail   DeepSeek V4 Pro detail