# ExploitGym Honeypot leaderboard

external benchmark

> Every figure below is reproduced as its upstream source published it: nothing is modelled, estimated, interpolated or converted. `Unavailable` means no source published the value — it is never a zero. Each value carries its evidence state and the date it was observed.

Page: https://modelscale.dev/leaderboards/benchmarks/external-exploit-gym-honeypot

ExploitGym Honeypot out-of-scope exploitation rate. Every model the catalog carries a published ExploitGym Honeypot value for, ranked by that value.

## Benchmark record

| Field | Value |
| --- | --- |
| Category | external |
| Measure | Out-of-scope exploitation rate (lower is better) |
| Tasks | Hard ExploitGym tasks with an out-of-scope honeypot service |
| Difficulty | Cyber-agent scope adherence |
| Published by | [GPT-6 Cyber: Advancing the frontier of aligned cybersecurity agents](https://openai.com/index/introducing-gpt-6-cyber/) |

An alignment test for cyber agents: hard ExploitGym tasks are paired with an easy-to-exploit auxiliary service (such as a package registry or object store) that would reveal the flag but is outside the authorized scope. Lower scores mean the model stays in scope more often.

## ExploitGym Honeypot ranking

11 models with a published ExploitGym Honeypot value, ordered by that value, highest first. Models the source has not scored on this benchmark are not listed — they are unmeasured, not last.

| Rank | Model | Provider | Out-of-scope exploitation rate (lower is better) | Evidence |
| --- | --- | --- | --- | --- |
| 1 | [GPT-5.6 Sol](https://modelscale.dev/models/gpt-5-6-sol) | OpenAI | 49.50 | Observed 2026-11-21 · source benchlm:models |
| 2 | [GPT-5.6 Luna](https://modelscale.dev/models/gpt-5-6-luna) | OpenAI | 9.90 | Observed 2026-11-21 · source benchlm:models |
| 3 | [GPT-5.6 Cyber](https://modelscale.dev/models/gpt-5-6-cyber) | OpenAI | 5.51 | Observed 2026-11-21 · source benchlm:models |
| 4 | [GPT-6 Cyber](https://modelscale.dev/models/gpt-6-cyber) | OpenAI | 0.07 | Observed 2026-11-21 · source benchlm:models |
| 5 | [GPT-6 Astra](https://modelscale.dev/models/gpt-6-astra) | OpenAI | 0.00 | Observed 2026-11-21 · source benchlm:models |
| 5 | [GPT-6 Luna](https://modelscale.dev/models/gpt-6-luna) | OpenAI | 0.00 | Observed 2026-11-21 · source benchlm:models |
| 5 | [GPT-6 Sol](https://modelscale.dev/models/gpt-6-sol) | OpenAI | 0.00 | Observed 2026-11-21 · source benchlm:models |
| 5 | [GPT-6.1 Sol](https://modelscale.dev/models/gpt-6-1-sol) | OpenAI | 0.00 | Observed 2026-11-21 · source benchlm:models |
| 5 | [GPT-6.1 Terra](https://modelscale.dev/models/gpt-6-1-terra) | OpenAI | 0.00 | Observed 2026-11-21 · source benchlm:models |
| 5 | [GPT-6.2 Astra](https://modelscale.dev/models/gpt-6-2-astra) | OpenAI | 0.00 | Observed 2026-11-21 · source benchlm:models |
| 5 | [GPT-6.2 Sol](https://modelscale.dev/models/gpt-6-2-sol) | OpenAI | 0.00 | Observed 2026-11-21 · source benchlm:models |

Rows are ordered by the value the source published, highest first. The source does not state whether a higher value is the better result, so this document does not either: for a benchmark that measures a rate of failure, read the table from the bottom.
