Head to head

GPT-5.6 Luna vs Llama 3.3 70B Instruct Turbo

On a blended workload of one million input tokens and three million output tokens, GPT-5.6 Luna is 1.1× cheaper than Llama 3.3 70B Instruct Turbo.

GPT-5.6 LunaLlama 3.3 70B Instruct Turbo
Provider OpenAI Together AI
Input / MTok $0.2 $1.04
Output / MTok $1.2 $1.04
Cache read / MTok $0.02
Context window 131,072
Blended 1M in + 3M out $3.80 $4.16
Status ga ga

GPT-5.6 Luna detail   Llama 3.3 70B Instruct Turbo detail