VariantBench: leaderboard
VariantBench evaluates AI agents on text-based genomics tasks involving genetic variant discovery, interpretation, and related QC, using realistic files and deterministic graders. The pinned metric is Endpoint pass rate (%), measuring the share of benchmark tasks successfully completed in the source-reported results table. Source: BenchmarkList’s attributed result table.
Metric: Endpoint pass rate (%). Source: benchmarklist.com. Saturation forecast: Around July 2028. 33 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 5 | 49.44 | #16 |
| 2 | GPT-6 Astra | 47.7 | #6 |
| 3 | Claude Opus 4.8 | 42.1 | #39 |
| 4 | GPT-5.6 Sol | 41.2 | #22 |
| 5 | Claude Sonnet 5 | 38.1 | #65 |
| 6 | Grok 4.6 | 38.1 | #29 |
| 7 | Grok 4.5 | 37.6 | #54 |
| 8 | Claude Opus 4.7 | 37 | #51 |
| 9 | Gemini 3.5 Flash | 35.3 | #66 |
| 10 | GPT-5.6 Terra | 34.7 | #40 |
| 11 | Gemini 3.1 Pro (Preview) | 34.5 | #64 |
| 12 | Kimi K3 | 32.8 | #35 |
| 13 | GPT-5.5 | 29.7 | #33 |
| 14 | GPT-5.6 Luna | 28.2 | #79 |
| 15 | Grok 4.3 | 19.5 | #204 |
No result here: #3 Claude Opus 5.5, #7 GPT-6.1 Sol, #9 Gemini 4 Argon.
Interactive version: theaggregate.ai/benchmark?slug=variantbench · How It Works · Data refreshed daily, snapshot 2026-11-05.