Will OpenAI's o4 get above 50% on humanity's last exam?
43
1kṀ7868
2027
18%
chance
3

Resolves N/A if there is no o4 model. o4 is defined as any compute setting on the o4 model. Something like deepresearch (which is based on o3/o4) would also resolve yes.

  • Update 2025-04-17 (PST) (AI summary of creator comment): o4 mini Exclusion Clarification

    • o4 is defined as any compute setting on the o4 model.

    • o4 mini is explicitly excluded from being considered as o4.

Technology
AI
Technical AI Timelines
OpenAI
AI Impacts
Get
Ṁ1,000
to start trading!

People are also trading

Top score on Humanity's Last Exam > 50% by 2029?
90% chance
Will an AI model surpasses o3's matharena.ai 88% Overall score by July 1, 2025?
58% chance
What will be the best AI performance on Humanity's Last Exam by December 31st 2025?
Will OpenAI's next major LLM (after GPT-4) surpass 70% accuracy on the GPQA benchmark?
75% chance
Will OpenAI's next major LLM (after GPT-4) surpass 74% accuracy on the GPQA benchmark?
86% chance
Humanity's Last Exam score in 2025?
50.8
When will OpenAI announce o4 (full)
Will "OpenAI o1" make the top fifty posts in LessWrong's 2024 Annual Review?
8% chance
Top score on Humanity's Last Exam > 50% by 2028?
69% chance
Top score on Humanity's Last Exam > 50% by 2027?
57% chance
Sort by:
bought Ṁ25 YES

They have to almost 4x the o4-mini score for this to happen, so definitely unlikely. However, given how much they were willing to spend on compute to get an unexpectedly high score on a similar high profile benchmark with o3 earlier it could happen, especially given a few months more of tinkering.

12% was simply a bit too low

o4 mini is not o4 btw

Humanity's last exam?

@ken https://agi.safe.ai/

People are also trading

Top score on Humanity's Last Exam > 50% by 2029?
+10% 1d90% chance
Will an AI model surpasses o3's matharena.ai 88% Overall score by July 1, 2025?
58% chance
What will be the best AI performance on Humanity's Last Exam by December 31st 2025?
Will OpenAI's next major LLM (after GPT-4) surpass 70% accuracy on the GPQA benchmark?
75% chance
Will OpenAI's next major LLM (after GPT-4) surpass 74% accuracy on the GPQA benchmark?
86% chance
Humanity's Last Exam score in 2025?
50.8
When will OpenAI announce o4 (full)
Will "OpenAI o1" make the top fifty posts in LessWrong's 2024 Annual Review?
8% chance
Top score on Humanity's Last Exam > 50% by 2028?
69% chance
Top score on Humanity's Last Exam > 50% by 2027?
57% chance
© Manifold Markets, Inc.•Terms•Privacy