What will be the best score (5/5 reliability) on ZeroBench by December 31st 2025?
3
แน742026
1D
1W
1M
ALL
21%
0 - 10
13%
11 - 20
8%
21 - 30
8%
31 - 40
8%
41 - 50
8%
51 - 60
8%
61 - 70
8%
71 - 80
8%
81 - 90
9%
91 - 100
This market will use the variant of the benchmark frozen one week after the initial release (following the public benchmark red-teaming stage to identify flawed/ambiguous questions).
The temperature used for the 5/5 reliability evaluation will be the default setting provided by each LLM API provider. In cases where this default is ambiguous to determine, we will default to a temperature of 0.7.
Get แน1,000 play money
Sort by:
As of March 28th 2025, Gemini 2.5 Pro is the new SotA: https://x.com/JRobertsAI/status/1905577784300183653
3% pass@1
5% pass@5
1% 5/5 reliability
Related questions
Related questions
What will be the best performance on FrontierMath by December 31st 2025?
Will any AI model score >80% on Epoch's Frontier Math Benchmark in 2025?
17% chance
Will an AI score over 80% on FrontierMath Benchmark in 2025
22% chance
What will be the best score on Cybench by December 31st 2025?
What will be the best normalized score achieved on the original 7 RE-Bench tasks by December 31st 2025?
What will be the best performance on SWE-bench Verified by December 31st 2025?
What will be the best performance on OSWorld by December 31st 2025?
Will any model get above human level on the Simple Bench benchmark before September 1st, 2025.
61% chance
What will be the best performance on EnigmaEval by December 31st 2025?
What will be the best AI performance on Humanity's Last Exam by December 31st 2025?