Head to head

Llama 3.1 8B Instant vs GPT-OSS 20B

On a blended workload of one million input tokens and three million output tokens, Llama 3.1 8B Instant is 2.2× cheaper than GPT-OSS 20B.

Llama 3.1 8B InstantGPT-OSS 20B
Provider Groq Together AI
Input / MTok $0.05 $0.05
Output / MTok $0.08 $0.2
Cache read / MTok
Context window 131,072 128,000
Blended 1M in + 3M out $0.29 $0.65
Status ga ga

Llama 3.1 8B Instant detail   GPT-OSS 20B detail