Expired70% confidenceFactTime unknown
The GPT-OSS 20B version activates only 3.6B parameters during inference and achieves performance close to O3-mini.
1
Sources
70%
Confidence
Medium-term (~90 days)
Relevance
6/1/2026
First Seen
Valid until: 8/30/2026(expired)
Sources
Related Entities
Related Claims
UnverifiedThe GPT-OSS 120B version activates only 5.1B parameters during inference and achieves performance close to O4-mini.92% similarUnverifiedGPT-OSS 20B performance falls between O3 mini and O4 mini.83% similarUnverifiedGPT-OSS 20B has 21B total parameters with 3.6B activated parameters during inference.81% similarUnverifiedGPT-5.5 still lags behind o4-mini in planning quality, and the best practice is to have o4-mini do the planning and GPT-5.5 handle the execution.79% similarUnverifiedGPT-4-level models typically achieve 30–80 TPS on consumer-grade APIs, while optimized small models or dedicated inference hardware can reach 200+ TPS.77% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/23234API
curl https://kongchang.com/api/v1/knowledge/claims/23234MCP
get_claim(id=23234)