scBench (LatchBio SingleCellBench)
195 verifiable single-cell RNA-seq analysis problems. Each pairs a data snapshot with a task and a deterministic grader.
Anthropic system cards call it LatchBio SingleCellBench with 195 problems, which matches the 195 problems in latchbio/scbench. Only six example problems are public, and the rest are withheld.
- Level
- Expert Specialists with years of training, often over hours of work
- Scored
- Checked automatically A program checks the result against an answer key, unit tests or the expected final state.
- Answers
- Agent task Many steps in a terminal, browser, desktop, app, game or robot.
Claude Opus 5.5 leads with 61.2%
Best score per model as labs report them in their own cards. Labs pick their own settings, so these compare less cleanly than independent runs.
- 1 Claude Opus 5.5 61.2%
- 2 Claude Opus 5 60.6%
- 3 Claude Sonnet 5.5 59.1%
- 4 Claude Opus 4.8 58.2%
- 5 Claude Haiku 5.5 56.2%
- 6 Claude Sonnet 5 56%
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Claude Opus 5.5 | 61.2% | Sep 2026 | Model card | |
| 2 | Claude Opus 5 | 60.6% | Jul 2026 | Model card | |
| 3 | Claude Sonnet 5.5 | 59.1% | Sep 2026 | Model card | |
| 4 | Claude Opus 4.8 | 58.2% | May 2026 | Model card | |
| 5 | Claude Haiku 5.5 | 56.2% | Oct 2026 | Model card | |
| 6 | Claude Sonnet 5 | 56% | Jun 2026 | Model card |
3 models have held number one since its release in Feb 2026
Lab claims count from the day the card came out, independent runs from the model’s release.
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Labs report it only in system cards
We read the headline results of 479 launch posts from 51 labs, and none of them list this benchmark. Some system cards do, below.
6 system cards report it, from 1 lab
Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.
| Date | Document | Score | As the document writes it | Section |
|---|---|---|---|---|
| 7 Oct 2026 | Claude Haiku 5.5 system card | 56.2 | LatchBio Bioinformatics, SingleCellBench, max effort, safeguards off (8.13.1, CB-1 evaluation in 2.2.2) | Bio and chem risk |
| 28 Sep 2026 | Claude Sonnet 5.5 system card | 59.1 | LatchBio SingleCellBench (8.17.2, Figure 8.17.A) | Capabilities |
| 22 Sep 2026 | Claude Opus 5.5 system card | 61.2 | LatchBio Bioinformatics, SingleCellBench, 195 problems, max effort (8.17.2) | Capabilities |
| 24 Jul 2026 | Claude Opus 5 system card | 60.6 | LatchBio Bioinformatics, SingleCellBench, 195 problems (8.17.2) | Capabilities |
| 30 Jun 2026 | Claude Sonnet 5 system card | 56 | LatchBio SingleCellBench, 195 problems, score 0.56 (8.14.2, Figure 8.14.6.A) | Capabilities |
| 28 May 2026 | Claude Opus 4.8 system card | 58.2 | LatchBio Bioinformatics, SingleCellBench, no extended thinking (8.16.3) | Capabilities |
More in Science and reasoning
- GPQA Diamond In use 198 graduate-level 'Google-proof' multiple-choice questions in biology, chemistry and physics.
Gemini 4 Argon 96.4% - Terminal-Bench-Science In use 70 scientific research-workflow tasks across five science areas, run in Harbor containers. Hidden tests give the resolution rate. GPT-6 Astra 68.1%
- FrontierScience In use Models solve physics, chemistry, and biology problems in an Olympiad track graded by short answer and a Research track graded by 10-point rubrics.
Intern-S2-397B 87% - BioMysteryBench In use Bioinformatics problems on anonymized real datasets that require real analysis with tools. Rubric-graded, with human-solvable and human-difficult splits. Claude Opus 5 90.1%
- HiPhO In use Models solve theory problems from 2024 and 2025 physics olympiad exams, some with diagrams. Official marking schemes grade answers and steps, mapped to medal levels. Gemini 3 Pro 78.3%
- LABBench2 In use 11 biology research sub-tasks (literature, figures, tables, patents, trials and more) with tool access. Macro-averaged accuracy. GPT-5.2 60.9%