ABC-Bench — leaderboard
Agentic Bio-Capabilities Benchmark for Biosecurity. Measures LLM agents on end-to-end biosecurity-relevant tasks such as liquid-handling robot code, DNA fragment design, and synthesis-screening evasion.
Metric: Overall pass@1 (self-reported). Source: benchmarklist.com. Status: saturation imminent. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Sonnet 4.5 | 63.2 |
| 2 | DeepSeek V3.2 | 50.1 |
| 3 | GPT-5 | 49.4 |
| 4 | GLM-4.7 | 40.1 |
| 5 | Gemini 2.5 Pro | 25 |
| 6 | Qwen 3 32B | 8.9 |
| 7 | Qwen 3 8B | 8.3 |
Interactive version: theaggregate.ai/benchmark?slug=abc-bench · How the rankings work · Data refreshed daily, snapshot 2026-07-22.