GPT-5.6 Sol Pro Aces Benchmark
jdjohnson · x · 2026-07-17
Shared reports indicate that GPT-5.6 Sol Pro scored 91/99 on prinzbench, nearly maxing out the benchmark.
Additional details:
- Two questions in prinzbench, worth a total of 6 points, remain unsolved by any model to date. Excluding these, GPT-5.6 Sol Pro actually solved 91 out of 93 available questions.
- OpenAI Pro series scores on this benchmark are as follows:
- GPT-5.4 Pro (Extended): 79/99
- GPT-5.5 Pro (Extended): 82/99
- GPT-5.6 Sol Pro: 91/99
- The original author noted that the benchmark was "saturated" just six months after its release in January 2026, highlighting the rapid pace of progress.
- They also mentioned that future OpenAI Pro models will no longer be compared using this benchmark in the same way.
More from Models
- Cohere's Invent-A-Dataset paper: zero-seed synthetic data beats frontier APIs on quality and diversity — idanbeck · 2026-10-27
- ARC Prize redesigns leaderboard: highest reasoning effort by default, CSV export — mhmazur · 2026-10-27
- Cohere: open-weight Qwen3.8-Flash-Next tried uploading probe file to Beijing cloud bucket in sandbox — wstone_bd · 2026-10-27
- Grok Bot auto-connects to X Money accounts without opt-in, user pulls funds — noahsolomon · 2026-10-27
- Model refuses wrong psychiatric advice in deliberately misleading test — FellMentKE · 2026-10-27
- Clinical System One v0.2: top clinical AI drops from 97% to 65% on harder cases — MaziyarPanahi · 2026-10-27