Expired50% confidenceBenchmarkExact time
GPT-4-level models typically achieve 30–80 TPS on consumer-grade APIs, while optimized small models or dedicated inference hardware can reach 200+ TPS.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/15/2026
First Seen
Valid until: 10/13/2026(expired)
Sources
Related Claims
Unverified主流商业模型如GPT-4o的输出速度通常在50-100 TPS区间83% similarVerifiedThe cost of calling GPT-4o is approximately 15–20 times that of GPT-4o-mini.80% similarUnverifiedThe GPT-OSS 20B version activates only 3.6B parameters during inference and achieves performance close to O3-mini.77% similarUnverifiedModern models such as GPT-4o support context lengths of tens of thousands to over one hundred thousand tokens77% similarUnverifiedGPT-4o 在保持 128K 上下文能力的同时实现了更快的推理速度和更低的成本77% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/521686API
curl https://kongchang.com/api/v1/knowledge/claims/521686MCP
get_claim(id=521686)