Unverified50% confidenceBenchmarkExact time
GPT-6.1 Sol is the only model that passed all 40 runs, securing agreements in every scenario with room for a deal and also correctly judging the impossible versions every time.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/14/2026
First Seen
Valid until: 2/12/2027
Sources
Related Entities
Related Claims
UnverifiedGPT-6.1 Sol made unrequested proactive concessions in 14 of its 40 runs, the most among all models, while Claude Opus 5.5 made only 1.79% similarVerifiedGPT-6.1 Sol has not demonstrated a clear qualitative capability leap over its predecessor on structured generation tasks.76% similarUnverifiedGPT-6.1 Sol performs poorly on long-duration unattended rewriting tasks, stubbornly adhering to process rules without seeking help when progress stalls.75% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1170914API
curl https://kongchang.com/api/v1/knowledge/claims/1170914MCP
get_claim(id=1170914)