Head to head

Safety GPT-OSS 20B vs DeepSeek V4 Flash (0731)

On a blended workload of one million input tokens and three million output tokens, Safety GPT-OSS 20B is line-ball with DeepSeek V4 Flash (0731).

Safety GPT-OSS 20BDeepSeek V4 Flash (0731)
Provider Groq Together AI
Input / MTok $0.075 $0.14
Output / MTok $0.3 $0.28
Cache read / MTok
Context window 131,072 1,000,000
Blended 1M in + 3M out $0.97 $0.98
Status preview ga

Safety GPT-OSS 20B detail   DeepSeek V4 Flash (0731) detail