BlogLLM Benchmark Scores as Published, Side by Side
FrontierScience: Results as Published
Every result the catalog holds for FrontierScience: 3 models from 1 developer, read from 1 page. Each number is as printed by the company named beside it; nothing is ranked or rescored. Models are listed by developer, newest first.
Area Science and MathMaintainer OpenAIIntroduced 2025Data version 2026103001, as of 2026-10-30
FrontierScience tests expert-level scientific work in two tracks. The Olympiad track holds problems at the level of the international physics, chemistry and biology olympiads with short checkable answers. The Research track holds open-ended, PhD-level sub-problems of real research questions, graded against rubrics by points rather than by a single answer. Release pages print the tracks separately.
Frontier Science Research
| Model | Value | Printed by | Setup as printed |
|---|---|---|---|
| GPT-5.4OpenAI, released 2026-03-05 | 33% | OpenAI, Introducing GPT-5.4checked 2026-10-15 | Effort: xhigh |
| GPT-5.4 ProOpenAI, released 2026-03-05 | 36.7% | OpenAI, Introducing GPT-5.4checked 2026-10-15 | Effort: xhigh |
| GPT-5.2OpenAI, released 2025-12-11 | 25.2% | OpenAI, Introducing GPT-5.4checked 2026-10-15 | Effort: xhigh |
Every number here is read from the same catalog file as the benchmark table, data version 2026103001, and is the number the table shows for that model and test. Where one page printed a model under several setups, each value is listed. The catalog is public in armanobosyan/sugra-llm-benchmarks; a wrong number can be reported from the table, with Report an error under the value, or as an issue in that repository.