Unverified75% confidenceFactTime unknown
GPT-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/s
1
Sources
75%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedWhen running GPT-OSS-20B on an RTX 5090 (32GB VRAM), GPU VRAM usage is approximately 53%, system memory usage is approximately 2.5GB, and response time is 1–2 seconds.80% similarUnverified单卡RTX 4090即可运行GPT-OSS 20B，单卡A100即可运行GPT-OSS 120B74% similarUnverified将模型权重从FP16压缩到INT8或INT4的量化技术，可使消费级GPU如RTX 4090在本地部署时达到30-80 TPS的实用速度73% similarUnverifiedWhen VRAM is sufficient and the model is fully resident on the GPU, an RTX 4090 running a Q4-quantized 27B model can typically achieve 15–30 tokens/s; once CPU offloading is triggered, the speed may drop sharply to 1–2 tokens/s or even lower.73% similarVerifiedThe consumer-grade GPU RTX 4090 has a VRAM ceiling of 24 GB.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/23209API
curl https://kongchang.com/api/v1/knowledge/claims/23209MCP
get_claim(id=23209)