Head to head

GPT-OSS 20B vs Qwen2.5 7B Instruct Turbo

On a blended workload of one million input tokens and three million output tokens, GPT-OSS 20B is 1.8× cheaper than Qwen2.5 7B Instruct Turbo.

GPT-OSS 20BQwen2.5 7B Instruct Turbo
Provider Together AI Together AI
Input / MTok $0.05 $0.3
Output / MTok $0.2 $0.3
Cache read / MTok
Context window 128,000 32,768
Blended 1M in + 3M out $0.65 $1.20
Status ga ga

GPT-OSS 20B detail   Qwen2.5 7B Instruct Turbo detail