What score will Anthropic's next Opus model achieve on GPQA?
1
Ṁ150Ṁ1resolved May 22
100%16%
< 80
16%
81 - 83%
16%
83 - 85%
16%
85 - 87%
18%
87 - 89%
16%
> 89%
https://www.theinformation.com/articles/anthropics-upcoming-models-will-think-think
The Information is reporting that Anthropic will release Claude Opus with reasoning in the next few weeks. Resolves when this model is released (regardless of what version number it has) and is benchmarked in GPQA. I will only be counting pass @ 1 and not the "parallel test time compute" numbers they additionally reported for 3.7 Sonnet.
If the model gets exactly on the edge of the range the higher of the two ranges it's in will resolve yes.
This question is managed and resolved by Manifold.
Market context
Get 
1,000 to start trading!
1,000People are also trading
Related questions
Will Anthropic’s next Sonnet model exceed 65% on terminal bench?
73% chance
Will Anthropic’s next Sonnet model exceed 83% on SWE-bench verified?
50% chance
In what year will AI achieve a score of 95% or higher on the GPQA benchmark?
5/25/27
[ACX 2026] What will be the highest score achieved on ARC-AGI-2 before 2027?
90.2
What will be the valuation of Anthropic in 2026? (M1000 subsidy)
Will Anthropic surpass OpenAI valuation in 2026?
29% chance
At what valuation will Anthropic IPO? (M1000 subsidy)
In what year will AI achieve a score of 95% or higher on the GSO benchmark?
4/8/29
What will be the next major event for Anthropic?