CategoryAgentic
MeasureResolution rate across 3 trials per task
Tasks70 expert-curated scientific research workflows
DifficultyFrontier scientific research workflows
Google-run scientific workflow evaluation using a sixfold verifier timeout to address verification timeouts.
Terminal-Bench-Science 0.1 (6x verifier timeout) ranking
1 model with a published Terminal-Bench-Science 0.1 (6x verifier timeout) value, ordered by that value, highest first. Models the source has not scored on this benchmark are not listed — they are unmeasured, not last.
Models ranked by their published Terminal-Bench-Science 0.1 with sixfold verifier timeout value| Rank | Model | Provider | Resolution rate across 3 trials per task |
|---|
| 1 |  Gemini 4 Argon | Google | 57.6 |
Evidence key: Observed
Rows are ordered by the value Gemini 4 Argon evaluation methodology published, highest first. The source does not state whether a higher value is the better result, so this page does not either: for a benchmark that measures a rate of failure, read the table from the bottom.
About Terminal-Bench-Science 0.1 (6x verifier timeout)
What the benchmark measures and how its value is published, from the source’s own record.
What is Terminal-Bench-Science 0.1 (6x verifier timeout)?
Terminal-Bench-Science 0.1 (6x verifier timeout) (Terminal-Bench-Science 0.1 with sixfold verifier timeout) is an agentic benchmark published by Gemini 4 Argon evaluation methodology. Google-run scientific workflow evaluation using a sixfold verifier timeout to address verification timeouts. Its evaluation set is described by the source as 70 expert-curated scientific research workflows. The source rates its difficulty as frontier scientific research workflows.
How is Terminal-Bench-Science 0.1 (6x verifier timeout) scored?
Each model's result is published by Gemini 4 Argon evaluation methodology as resolution rate across 3 trials per task. ModelScale prints that value to 1 decimal place, exactly as published, and never converts or rescales it. This page ranks the 1 model that carry a published value, highest value first; a model the source never scored is left out rather than ranked at zero. The source does not state whether a higher value is the better result, so neither does this page.
ModelScale