SciCode-Verified: How Benchmark Defects Underestimated the Scientific-Coding Ability of Language Models

• URL: http://arxiv.org/abs/2608.04975v1
• Date: Wed, 05 Aug 2026 15:45:55 GMT
• Title: SciCode-Verified: How Benchmark Defects Underestimated the Scientific-Coding Ability of Language Models
• Abstract summary: SciCode is the standard measure of the scientific-coding ability of language models.<n>It is a component of the Artificial Analysis Intelligence Index.<n>State-of-the-art models are far more proficient in scientific coding than SciCode has suggested.
• Score: 1.9873319610438742
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: SciCode is the standard measure of the scientific-coding ability of language models: research-level problems that demand both frontier scientific theory and its implementation as working numerical code. It is a component of the Artificial Analysis Intelligence Index and a standing evaluation in government and national-laboratory suites. Yet its scores have recently plateaued: the strongest 2026 models cluster tightly around 60\% subproblem accuracy, and a successor model ties its predecessor. We trace this stagnation to defects in the benchmark itself. A per-problem, domain-expert audit of all 65 test problems uncovers 263 defects; 192 of them, spread across 91\% of the main problems, cause correct, instruction-following solutions to be wrongly rejected---through non-reproducible gold answers, over-tight tolerances, or self-contradictory specifications. Critically, 78\% of these score-suppressing defects require specialized physics or mathematics knowledge to detect, not mere clerical proofreading. We corrected every confirmable defect to produce SciCode-Verified. The corrections add only the specifications a well-posed problem requires, repair grading, and tighten the tests that were too lenient; every change is recorded with its justification and independently re-checked by a second domain expert. We re-evaluate twelve frontier model snapshots on the corrected benchmark and find a substantial recovery: subproblem accuracy rises from 45--60\% to 84--98\%, and main-problem accuracy from 9--27\% to 69--92\%. State-of-the-art models are far more proficient in scientific coding than SciCode has suggested---the bottleneck was not model capability, but the quality of the evaluation instrument. We release SciCode-Verified with its complete audit trail as the corrected public standard.

Related papers

• How Good Are Frontier Models at Physics? Expert Re-Grading Reveals Broken Evaluations and Near-Saturation of Leading Benchmarks [60.853503196898906]
  Low reported scores on leading physics benchmarks suggest frontier language models struggle with advanced physics.<n>We revisit these reported findings by evaluating frontier models on six widely used physics benchmarks and auditing them with experts.
  arXiv  Detail & Related papers  (2026-09-11T16:06:50Z)
• IdeaAMBIG: Benchmarking Implementation-Critical Gaps in Research-Idea Specifications [52.570108663867046]
  A research idea may be novel, coherent, and scientifically plausible, yet its proposed method may remain insufficiently specified for faithful implementation.<n>We study the codification readiness of implementation-facing research-method specifications, defined by whether they provide sufficient methodological information for a competent implementer or coding agent to construct the intended method without unsupported assumptions.<n>We introduce IdeaAMBIG, a benchmark of 660 evidence-grounded instances: 163 real-world gaps from reports and GitHub issues, and 497 controlled synthetic gaps injected into codification-ready references.
  arXiv  Detail & Related papers  (2026-09-09T17:59:04Z)
• Agentic Proving for Program Verification [44.663012714194025]
  Agentic systems have emerged as state-of-the-art approaches for automated theorem proving in formal mathematics.<n>We evaluate Claude Code in an agentic proving framework on CLEVER, a Lean 4 benchmark for verifiable code generation.
  arXiv  Detail & Related papers  (2026-05-22T15:41:27Z)
• Correct Answers from Sound Reasoning: Verifiable Process Supervision for Language Models [94.68358825189738]
  We propose a post-training framework for verifiable domains that jointly optimize prediction accuracy and reasoning quality.<n>We evaluate VPS on chess, a controlled testbed where reasoning steps can be deterministically verified against engine signals.<n>VPS preserves accuracy while significantly improving reasoning quality, reducing win-rate error by up to 30% and restoring consistency to near saturation.
  arXiv  Detail & Related papers  (2026-04-03T15:19:46Z)
• A Judge Agent Closes the Reliability Gap in AI-Generated Scientific Simulation [1.7259824817932294]
  Large language models can generate scientific simulation code, but the generated code silently fails on most non-textbook problems.<n>We show that classical mathematical validation can be fully automated by a Judge Agent.<n>72 blinded tasks submitted by 12 independent scientists yield an 89% success rate with automated error bounds, versus 53% without the Judge.
  arXiv  Detail & Related papers  (2026-03-26T16:03:52Z)
• HLE-Verified: A Systematic Verification and Structured Revision of Humanity's Last Exam [63.84155758655084]
  Humanity's Last Exam (HLE) has become a widely used benchmark for evaluating frontier large language models.<n>We introduce HLE-Verified, a verified and revised version of HLE with a transparent verification protocol and fine-grained error taxonomy.<n>We evaluate seven state-of-the-art language models on HLE and HLE-Verified, observing an average absolute accuracy gain of 7--10 percentage points.
  arXiv  Detail & Related papers  (2026-02-15T02:50:15Z)
• SciCoQA: Quality Assurance for Scientific Paper--Code Alignment [53.70401063640645]
  We present SciCoQA, a dataset for detecting discrepancies between scientific publications and theirs.<n>Our dataset consists of 611 paper-code discrepancies (81 real, 530 synthetic), spanning diverse computational science disciplines.<n>The best performing model in our evaluation, GPT-5, can only detect 45.7% of real-world paper-code discrepancies.
  arXiv  Detail & Related papers  (2026-01-19T10:04:33Z)
• MSC-180: A Benchmark for Automated Formal Theorem Proving from Mathematical Subject Classification [21.9173105378467]
  Current large language model (LLM)-based theorem provers suffer from limitations such as restricted domain coverage and weak generalization in mathematical reasoning.<n>We propose MSC-180, a benchmark for evaluation based on the MSC 2020 mathematical subject classification.<n>It comprises 180 formal verification problems, 3 advanced problems from each of 60 mathematical branches, spanning from undergraduate to graduate levels.
  arXiv  Detail & Related papers  (2025-12-20T07:39:19Z)
• LLM Critics Help Catch Bugs in Mathematics: Towards a Better Mathematical Verifier with Natural Language Feedback [71.95402654982095]
  We propose Math-Minos, a natural language feedback-enhanced verifier. Our experiments reveal that a small set of natural language feedback can significantly boost the performance of the verifier.
  arXiv  Detail & Related papers  (2024-06-20T06:42:27Z)
• Evaluating Mathematical Reasoning Beyond Accuracy [50.09931172314218]
  We introduce ReasonEval, a new methodology for evaluating the quality of reasoning steps.<n>We show that ReasonEval consistently outperforms baseline methods in the meta-evaluation datasets.<n>We observe that ReasonEval can play a significant role in data selection.
  arXiv  Detail & Related papers  (2024-04-08T17:18:04Z)
• Assessing the Efficacy of Grammar Error Correction: A Human Evaluation Approach in the Japanese Context [10.047123247001714]
  We evaluate the performance of the state-of-the-art sequence tagging grammar error detection and correction model (SeqTagger) With an automatic annotation toolkit, ERRANT, we first evaluated SeqTagger's performance on error correction with human expert correction as the benchmark. Results indicated a precision of 63.66% and a recall of 20.19% for error correction in the full dataset.
  arXiv  Detail & Related papers  (2024-02-28T06:43:43Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.