ExploitBench - Internal Port
Agents develop working exploits for recently disclosed vulnerabilities to limit contamination. Scored as arbitrary code execution success rate.
OpenAI internal port of ExploitBench with June to August 2026 vulnerabilities.
- Level
- Research frontier Researchers, on open problems or tasks that take days
- 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.
GPT-6.1 Terra leads with 24.5%
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 GPT-6.1 Terra 24.5%
- 2 GPT-6.1 Sol 21.5%
- 3 GPT-6 Sol 5.50%
- 4 GPT-6 Luna 0%
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-6.1 Terra | 24.5% | Oct 2026 | Model card | |
| 2 | GPT-6.1 Sol | 21.5% | Sep 2026 | Model card | |
| 3 | GPT-6 Sol | 5.50% | Sep 2026 | Model card | |
| 4 | GPT-6 Luna | 0% | Sep 2026 | Model card |
3 models have held number one since its release in Sep 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.
5 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 |
|---|---|---|---|---|
| 26 Oct 2026 | GPT-6.1 Terra system card addendum | 24.5 | ExploitBench - Internal Port (June–August 2026), arbitrary code-execution success rate (9.0.2.2) | Cyber risk |
| 29 Sep 2026 | GPT-6.1 Sol system card addendum | 21.5 | ExploitBench - Internal Port (June–August 2026), arbitrary code-execution success rate (9.1.2.2) | Cyber risk |
| 22 Sep 2026 | GPT-6 Sol system card | 5.5 | ExploitBench - Internal Port (June–August 2026), maximum success rate (A.8.1.2.2) | Cyber risk |
| 22 Sep 2026 | GPT-6 Luna system card | 0 | ExploitBench - Internal Port (June–August 2026), text says no challenge completed (A.8.1.2.2) | Cyber risk |
| 3 Sep 2026 | GPT-6 Astra system card | ExploitBench - Internal Port (June–August 2026), success rate vs output tokens, chart without printed numbers (Figure 47) | Cyber risk |
More in Cybersecurity
- ExploitGym In use 869 real-vulnerability exploit-development tasks (userspace, V8, Linux kernel). Success means capturing a flag through the intended vulnerability. GPT-6 Astra 42.4%
- ExploitBench In use 41 recent V8 JavaScript-engine N-day vulnerabilities. Progress measured by 16 grader-verified capability flags across five tiers. GPT-6 Astra 100%
- CWE-bench In use Agents audit a repository and write a patch for a security flaw, without being told where the bug is. Covers six programming languages.
Gemini 4 Carbon 71% - Real-world vulnerability discovery In use Agents search popular open-source codebases to rediscover about 1,200 recent confirmed vulnerabilities across many languages. Reports share discovered. Gemini 4 Argon 86%
- Artificial Analysis Cyber Index In use Agents do enterprise cyber defense tasks from three benchmarks, CWE-Bench-AA, DeepsecBench-AA and CyberGym-E2E-AA, combined into one success rate.
Mistral Large 4 (Le Chonk, open weights) 51% - Big Sleep Evaluation (Google) In use Agents search large real codebases such as Chrome and Safari for critical, hard to find vulnerabilities. Scored by pass@1 success rate, run without safety guardrails. Gemini 3.5 Flash Cyber 72%