#clockbench search results
Wild jump—GPT-5 Pro at 45% on the 10-clock sample vs ~13% best official. Anyone else run ClockBench? Drop your % + model + UI/API + prompt. Also want an o3-pro vs GPT-5 Pro head-to-head. Props @alek_safar 🙌 #ClockBench #AIEvals #LLM
Super interesting, Joe. Could variance/seed be doing work here? Would love to see: zero-shot vs few-shot, UI vs API, and any prompt template you used. Community runs welcome—share your CSV + script if you’ve got it. #Reproducibility #AIEvals #ClockBench
🕒 Introducing ClockBench where humans still crush AI at a task as simple as telling the time. 🌐 Explore more here 👉 clockbench.ai @PMinervini @aryopg @rohit_saxena #ClockBench #AIResearch #VisualReasoning #MultimodalAI #Benchmarking
Something went wrong.
Something went wrong.
United States Trends
- 1. Sonny Gray 1,300 posts
- 2. #GMMTV2026 4.16M posts
- 3. Thankful 49.9K posts
- 4. #csm221 2,454 posts
- 5. #OurCosmicClue_Wooyoung 24.3K posts
- 6. Gone in 60 1,101 posts
- 7. National Treasure 3,759 posts
- 8. Happy Thanksgiving 18.4K posts
- 9. Mark Kelly 249K posts
- 10. Mainz Biomed N/A
- 11. MILKLOVE BORN TO SHINE 712K posts
- 12. Hegseth 119K posts
- 13. #LUNÉSelcaDay 2,749 posts
- 14. #YouManiacSeries 102K posts
- 15. Good Tuesday 39.4K posts
- 16. Ghost Rider 1,284 posts
- 17. Lord of War N/A
- 18. Raising Arizona N/A
- 19. Alan Dershowitz 5,425 posts
- 20. Taco Tuesday 14.3K posts