Unverified50% confidenceBenchmarkExact time
OpenAI's GPT-OSS 20B achieved a measured speed of over 21 tokens/second, nearly twice that of the dense 12B model.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/19/2026
First Seen
Valid until: 12/18/2026
Sources
Related Entities
Related Claims
Unverified测试者用双4090（48GB显存）工作站运行20.7B稠密模型，每秒生成27.6个Token，仅比矮星略快，但参数量小10倍以上76% similarUnverifiedGPT-OSS 20B推理仅需16G显存，微调仅需24G显存75% similarUnverifiedGPT-OSS-20B BF16 loading was reduced from 545 seconds to 70 seconds (7.8× speedup), and Qwen3.5-397B achieved a 1.93× to 2.3× improvement.75% similarUnverifiedGPT-OSS 20B在RTX 3090上可达约40 Token/s的生成速度，在RTX 5090上能达到200 Token/s74% similarUnverifiedA 2B model requires approximately 4–10 TFLOPs of computation per inference, taking 10–50 milliseconds on an A100 GPU, while a 70B model requires 140–350 TFLOPs and can take up to several seconds.74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/933720API
curl https://kongchang.com/api/v1/knowledge/claims/933720MCP
get_claim(id=933720)