LmCast :: Stay tuned in

How good are frontier models at physics?

Recorded: Sept. 17, 2026, 12:28 a.m.

Original Summarized

[2609.13009] How Good Are Frontier Models at Physics? Expert Re-Grading Reveals Broken Evaluations and Near-Saturation of Leading Benchmarks

Skip to main content

Search

Submit
Donate

Log in

Search arXiv

Press Enter to search · Advanced search

Computer Science > Artificial Intelligence

arXiv:2609.13009 (cs)

[Submitted on 11 Sep 2026]
Title:How Good Are Frontier Models at Physics? Expert Re-Grading Reveals Broken Evaluations and Near-Saturation of Leading Benchmarks
Authors:Ali Ansari, Haoran Sun, Andy Zeyi Liu, Mark Jabbour, Yongshan Ding, Steven Girvin, Yu He, Sohrab Ismail-Beigi, Aleksander Kubica, Owen D. Miller, Corey O'Hern, Vidvuds Ozolins, David Poland, A. Douglas Stone, Frank C. van den Bosch, Logan Wright, Navid Akbari, Santanu Antu, Kangle Cai, Andrew Calabrese-Day, Mateo Cárdenes Wuttig, Meng Cheng, Barry T. Chiang, Ali Ghorashi, Shouzhen Gu, Haoyang Huang, Zhibo Kang, Lukas Kienesberger, Hantian Liu, Charles Lomba, Zhongling Lu, Wenchao Ma, Rohin E. McIntosh, Evan McKinney, Ivan Rojkov, Xulei Sun, Yarone Meir Tokayer, Naveen Balaji Umasankar, Mira Varma, Leda Wang, Qimin Wang, Tyler Wang, Haoyu Wei, Jinming Yang, Jinchen Zhao, Sherlock Tingrui Zhao, Qinyuan Zheng, Jay S. Zou, Lucas Baker, Arman Cohan, John Sous View a PDF of the paper titled How Good Are Frontier Models at Physics? Expert Re-Grading Reveals Broken Evaluations and Near-Saturation of Leading Benchmarks, by Ali Ansari and 50 other authors
View PDF
HTML (experimental)

Abstract:Low reported scores on leading physics benchmarks, including those featured in the Artificial Analysis Intelligence Index (2026), suggest that frontier language models still struggle with advanced physics, a demanding test of their scientific reasoning and quantitative problem-solving abilities. Yet this impression does not always align with domain experts' experiences using these models in their work. We revisit these reported findings by evaluating frontier models on six widely used physics benchmarks and auditing them with experts, focusing on text-only problems with verifiable final answers. For each subfield of physics, faculty and graduate researchers with relevant expertise carefully review problem statements, reference solutions, and model responses to distinguish genuine model errors from grader errors, incorrect reference solutions, and ambiguous or underspecified questions. Most audited cases initially evaluated as incorrect reflect these benchmarking issues rather than errors in the models' physics reasoning. We then ask experts to address these benchmarking issues by correcting erroneous reference solutions and repairing or excluding flawed questions. We find that GPT-5.6-Sol's measured mean@4 rises from 47.3% to 78.7% on HLE-Physics and from 61.0% to 87.2% on CMT-Benchmark, while its corrected pass@4 reaches 94.4% on the 54 retained CritPt challenges. Corrected scores are computed on the retained evaluation subsets following expert review. Scores on the audited subsets of UGPhysics, PRISM-Physics, and PHYBench also rise substantially after correction. These findings suggest that current benchmarks substantially understate frontier models' ability to solve well-posed physics problems. Near-saturation on these closed-ended tasks highlights the need for more demanding, expert-validated evaluations.

Subjects:

Artificial Intelligence (cs.AI)

Cite as:
arXiv:2609.13009 [cs.AI]

 
(or
arXiv:2609.13009v1 [cs.AI] for this version)

 
https://doi.org/10.48550/arXiv.2609.13009

Focus to learn more

arXiv-issued DOI via DataCite (pending registration)

Submission history From: Ali Ansari [view email] [v1]
Fri, 11 Sep 2026 16:06:50 UTC (178 KB)

Full-text links:
Access Paper:

View a PDF of the paper titled How Good Are Frontier Models at Physics? Expert Re-Grading Reveals Broken Evaluations and Near-Saturation of Leading Benchmarks, by Ali Ansari and 50 other authorsView PDFHTML (experimental)TeX Source

view license


Current browse context:
cs.AI

< prev

  |  
next >

new
|
recent
| 2026-09

Change to browse by:

cs

References & Citations

NASA ADSGoogle Scholar
Semantic Scholar

export BibTeX citation
Loading...

BibTeX formatted citation
×

loading...

Data provided by:

Bookmark

Bibliographic Tools

Bibliographic and Citation Tools

Bibliographic Explorer Toggle

Bibliographic Explorer (What is the Explorer?)

Connected Papers Toggle

Connected Papers (What is Connected Papers?)

Litmaps Toggle

Litmaps (What is Litmaps?)

scite.ai Toggle

scite Smart Citations (What are Smart Citations?)

Code, Data, Media

Code, Data and Media Associated with this Article

alphaXiv Toggle

alphaXiv (What is alphaXiv?)

Links to Code Toggle

CatalyzeX Code Finder for Papers (What is CatalyzeX?)

DagsHub Toggle

DagsHub (What is DagsHub?)

GotitPub Toggle

Gotit.pub (What is GotitPub?)

Huggingface Toggle

Hugging Face (What is Huggingface?)

ScienceCast Toggle

ScienceCast (What is ScienceCast?)

Demos

Demos

Replicate Toggle

Replicate (What is Replicate?)

Spaces Toggle

Hugging Face Spaces (What is Spaces?)

Spaces Toggle

TXYZ.AI (What is TXYZ.AI?)

Related Papers

Recommenders and Search Tools

Link to Influence Flower

Influence Flower (What are Influence Flowers?)

Core recommender toggle

CORE Recommender (What is CORE?)

Author
Venue
Institution
Topic

About arXivLabs

arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? |
Disable MathJax (What is MathJax?)

We gratefully acknowledge support from
our major funders,
member institutions, ,
and all contributors.

About
·
Help
·
Contact
·
Subscribe
·
Copyright
·
Privacy
·
Accessibility
·
Operational Status (opens in new tab)

Major funding support from

The study investigates the actual capabilities of frontier language models when applied to physics, addressing the discrepancy between reported benchmark scores and expert assessments. The investigation was prompted by low performance scores observed on leading physics benchmarks, including those featured in the Artificial Analysis Intelligence Index 2026, which suggested that these models struggle with the demanding requirements of scientific reasoning and quantitative problem-solving. To validate these initial findings, the researchers reevaluated frontier models by assessing them on six widely used physics benchmarks and employing domain experts to audit the results.

The auditing process involved having faculty and graduate researchers with relevant expertise meticulously review the problem statements, the provided reference solutions, and the models' responses. The goal of this expert review was to differentiate between genuine errors made by the language models, errors in the benchmarking process itself, incorrect reference solutions, or issues arising from ambiguous or underspecified questions. A key observation from this process was that most instances initially flagged as incorrect in the benchmarking reflected flaws in the evaluation setup rather than actual deficits in the models' physics reasoning abilities. To refine the assessment, experts were subsequently engaged to address these identified benchmarking issues by correcting erroneous reference solutions and either repairing or excluding questions deemed fundamentally flawed.

Following this expert-validated correction process, substantial performance improvements were documented for the models. Specifically, the measured mean@4 scores for GPT-5.6-Sol increased significantly, rising from 47.3% to 78.7% on the HLE-Physics benchmark and from 61.0% to 87.2% on the CMT-Benchmark. Furthermore, after applying the expert corrections, the corrected pass@4 metric reached 94.4% across the fifty-four retained CritPt challenges. Similar substantial increases were observed when scores were corrected for the audited subsets of UGPhysics, PRISM-Physics, and PHYBench. These empirical results collectively suggest that existing benchmarks significantly underestimate the actual capacity of frontier models to solve well-posed physics problems. Consequently, the findings highlight an urgent necessity for the development of more rigorous, expert-validated evaluation methods to accurately gauge the scientific reasoning capabilities of these advanced models, especially given the observed near-saturation of performance on these closed-ended tasks.