Science's Last Exam
Science's Last Exam is Can AI do research, end to end? Science's Last Exam defines and measures whether an AI agent can carry a research goal around the whole loop of scientific discovery, with the real world as the judge, and diagnoses…. It is ranked #104 on the AI Radar, in AI research, first seen 20 h ago and shared in 6 posts (38.9k views).
Science's Last Exam Science’s Last Exam Sign in Manage account Sign out A benchmark for scientific agents Science’s Last Exam Can AI do research, end to end? Watch video Contribute Blog GitHub Dataset 113 public tasks 6 capabilities 1 end-to-end 9 disciplines 01 Hypothesis Discovery ↗ ✓ Post-cutoff studies 02 Protocol Planning ↗ ✓ Reference protocol 03 Scientific Computing ↗ ✓ Original code run 04 Scientific Software Operation ↗ ✓ Saved project file 05 Wet-lab Execution (In construction) ↗ ✓ Instrument readout 06 Experiment Observation ↗ ✓ Events in the video 07 End-to-end research (To be expanded) ✓ Held-out checks One research goal. Nine…
What people said about Science's Last Exam on X
Can AI agents do scientific research on their own? Science's Last Exam is a new benchmark measure this. It's built from real research tasks across nine scientific fields. It tests whether an agent can form hypotheses, plan protocols, run scientific code and software, and read what happens in experiment videos.…
— @omarsar0, 20 h ago · 125 likes · see the post
Here we go! Science's last exam. A benchmark for scientific research: statistics, physics, astronomy, chemistry, earth science, biology, engineering, materials science, and... medicine! Models are scored on 5 domains: hypothesis discovery, protocol planning, scientific computing, software operation and experiment…
— @DrSamuelBHume, 19 h ago · 30 likes · see the post
Lean gave AI math a grader nobody can argue with, and empirical science is now getting its own version of that idea. Science’s Last Exam (SLE), a benchmark, grades research agents on artifacts that can be opened and re-run, with checks that domain experts seal when each task is built. Led by the US nonprofit AItonomy…
— @rohanpaul_ai, 18 h ago · 26 likes · see the post
Alternatives to Science's Last Exam
- flashprove — Further progress in provable inference will likely be an important part of improving trust in AI deployment. To that…
- moe-blogpost — A hands-on guide to training Mixture-of-Experts models
- liberai.com — Technical blog
- science-last-exam — Science's Last Exam. Contribute to Gen-Verse/Science-Last-Exam development by creating an account on GitHub.
- www-cdn.anthropic.com — Technical report
- depth-hybrid-attention — Project overview & code
Science's Last Exam in numbers
- Rank on the AI Radar: #104 of 10389
- Shared in 6 posts by 6 accounts: @omarsar0, @DrSamuelBHume, @rohanpaul_ai, @undefinedKi
- 38.9k views on those posts
- First seen 20 h ago, last shared 5 h ago
- Pricing seen by AGK Intelligence: not stated
- Market: AI research
FAQ
What is Science's Last Exam?
Can AI do research, end to end? Science's Last Exam defines and measures whether an AI agent can carry a research goal around the whole loop of scientific discovery, with the real world as the judge, and diagnoses… It was first shared on X 20 h ago and is ranked #104 on the AI Radar.
Is Science's Last Exam free?
Pricing is not stated on the page we read.
Who shared Science's Last Exam?
6 accounts on X, including @omarsar0, @DrSamuelBHume, @rohanpaul_ai, in 6 posts totalling 38.9k views.
Every post is read and classified by AGK Intelligence: what it is, which market it belongs to, and whether the link is a real tool. 65.7k posts from 5k X accounts over the last 14 days, 10.4k tools, 19 markets. New posts are collected several times an hour and the full ranking is recomputed every hour — last update 2026-11-06 13:30 UTC. Full method.