FrontierScience Research
We show this table for reference; we do not rank on it.
A research-focused FrontierScience evaluation variant for scientific investigation and problem solving.
Benchmark score on FrontierScience Research — November 13, 2026
We compile the FrontierScience Research rows from provider self-reports and secondary reports. Apodex 1.1 leads the table at 63.3%, followed by Apodex 1.1 Mini (51.7%) and Sage-A1.5 (48.3%). We do not use these results to rank models overall.
Apodex 1.1
Apodex
Reported by ApodexApodex 1.1 Mini
Apodex
Reported by ApodexSage-A1.5
InternScience
Reported by InternScience5 modelsKnowledgeCurrentDisplay onlyUpdated November 13, 2026
Benchmark score table (5 models)
None of the 5 results on this page has a documented run by the benchmark or an independent evaluator: 4 reported by the labs that built the models and 1 reported in another publisher’s document.
| Rank | Model / configuration | Score | Who ran it | Parameters (B) | Open / closed |
|---|---|---|---|---|---|
| 1 | Apodex 1.1ApodexReported by Apodex | 63.3% | Reported by Apodex | Not reported | Closed |
| 2 | Apodex 1.1 MiniApodexReported by Apodex | 51.7% | Reported by Apodex | Not reported | Open |
| 3 | Sage-A1.5InternScienceReported by InternScience | 48.3% | Reported by InternScience | Not reported | Open |
| 4 | GPT-5.4 ProOpenAIReported by Meta AI (comparison table) | 36.7% | Reported by Meta AI (comparison table) | Not reported | Closed |
| 5 | Agents-A1-4BInternScienceReported by InternScience | 33.3% | Reported by InternScience | Not reported | Open |
- Lab-reported
- Benchmark-run
- Independent
Results reported by model developersUnranked
Labs ran these themselves, with a metric, split or setting the table above does not use. They take no rank here.
| Model | Reported value | What was measured | Against the table above | Who ran it |
|---|---|---|---|---|
| Apodex 1.1Apodex | 55.0%Pass rate (7 of 10 rubric points) | ReAct single agentThe stored row uses Agent Team | Not comparable to the scores above | Reported by Apodex |
| Apodex 1.1 MiniApodex | 45.0%Pass rate (7 of 10 rubric points) | ReAct single agentThe stored row uses Agent Team | Not comparable to the scores above | Reported by Apodex |
A lab’s own result is shown for reference. It never takes a rank on this board.
How we label sourcesAmong the reported FrontierScience Research rows, Apodex 1.1 is first at 63.3%. The third row is 15.0 points behind. The broader top-10 range is 30.0 points, so the table still separates the published systems.
5 models have been evaluated on FrontierScience Research. The benchmark falls in the Knowledge category. FrontierScience Research is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About FrontierScience Research
Year
2026
Tasks
Scientific research problems
Format
Research evaluation
Difficulty
Frontier scientific research
Meta uses FrontierScience Research in its Contemplating-mode comparison table as a distinct scientific research variant. BenchLM stores it as a display-only frontier science reference.
Freshness and provenance
Version
FrontierScience Research 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
Questions
What does FrontierScience Research measure?
A research-focused FrontierScience evaluation variant for scientific investigation and problem solving.
Which model scores highest on FrontierScience Research?
Apodex 1.1 by Apodex currently leads with a score of 63.3% on FrontierScience Research.
How many models are evaluated on FrontierScience Research?
5 AI models have published results on FrontierScience Research in the BenchLM catalog.
Compare top models on FrontierScience Research
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 7,000+ subscribers.
One email each week. Unsubscribe anytime.