Unverified50% confidenceBenchmarkExact time
In a 2030 World Cup partition question, gpt-oss-20b gave six exhaustive mutually exclusive outcomes that summed to only 0.36.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/11/2026
First Seen
Valid until: 1/9/2027
Sources
Related Entities
Related Claims
UnverifiedIn an unknown-rules mini-game test designed specifically for models, GPT-6 scored 99 points, approaching a perfect score, while the highest human score was only in the 30s and other models scored around 40.63% similarUnverified21.2% is a relative reduction rather than an absolute one, meaning the new version eliminates approximately one-fifth of the failures relative to the earlier version's failure rate baseline.63% similarUnverifiedAccording to OpenAI's early 2024 technical report, by annotating sycophantic behavior in RLHF data and assigning negative rewards, the unconditional acquiescence rate of subsequent GPT-4 iterations declined.63% similarUnverified2024 年初最先进的模型在 OS World Verified 测试上通常连 20% 都拿不到62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/999215API
curl https://kongchang.com/api/v1/knowledge/claims/999215MCP
get_claim(id=999215)