# Atomic Vulnerability Research and Exploitation (Atomic vulnerability research)

> Irregular's domain-level evaluation of vulnerability research and exploitation capability.

Canonical page: https://benchlm.ai/benchmarks/atomicvulnerabilityresearch

- Category: [Agentic](/agentic)
- Last updated: September 27, 2026

## About Atomic vulnerability research

- Year: 2026
- Tasks: Atomic cyber tasks
- Format: Domain average
- Difficulty: Vulnerability research and exploitation
- Paper: [Assessing GPT-5.6 Sol](https://www.irregular.com/research/assessing-gpt-5.6-sol)

This Atomic domain average isolates vulnerability research and exploitation. BenchLM stores it as an independent, display-only signal.

Atomic vulnerability research is currently displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Leaderboard (3 models)

| Rank | Model | Creator | Score |
|------|-------|---------|-------|
| 1 | [GPT-6 Astra](/models/gpt-6-astra) | OpenAI | 100% |
| 2 | [GPT-5.5](/models/gpt-5-5) | OpenAI | 92% |
| 3 | [GPT-5.6 Sol](/models/gpt-5-6-sol) | OpenAI | 91% |

## FAQ

### What does Atomic vulnerability research measure?

Irregular's domain-level evaluation of vulnerability research and exploitation capability.

### Which model scores highest on Atomic vulnerability research?

GPT-6 Astra by OpenAI currently leads with a score of 100% on Atomic vulnerability research.

### How many models are evaluated on Atomic vulnerability research?

3 AI models have been evaluated on Atomic vulnerability research on BenchLM.

### Does Atomic vulnerability research affect BenchLM's overall score?

Not directly. Atomic vulnerability research is still displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Compare Top Models on Atomic vulnerability research

- [GPT-6 Astra vs GPT-5.5](/compare/gpt-5-5-vs-gpt-6-astra)
- [GPT-5.5 vs GPT-5.6 Sol](/compare/gpt-5-5-vs-gpt-5-6-sol)