# EBR-bench → ExploitBench

Calibration · https://benchgap.net/calibration/ebrbench/current/to/exploitbench/current · data retrieved 2026-11-18

ExploitBench is estimated from EBR-bench with a inverse Michaelis–Menten curve fitted on 5 models measured on both: y = 3.32881·(x − 0.3317) / (0.3317 + 0.6428 − x), R² = 0.98, cross-validated error 6.4 pp. It is used for 17 estimates.

| Estimated model | EBR-bench | ExploitBench | Source |
|---|---|---|---|
| [Claude Fable 5.1](https://benchgap.net/model/claude-fable-5-1.md) | 57.1% | 100.0% | estimated ± 6.4 pp, low confidence |
| [Claude Fable 5.2](https://benchgap.net/model/claude-fable-5-2.md) | 90.5% | 100.0% | estimated ± 6.4 pp, low confidence |
| [Claude Haiku 5.5](https://benchgap.net/model/claude-haiku-5-5.md) | 43.8% | 65.9% | estimated ± 6.4 pp, low confidence |
| [Claude Fable 5](https://benchgap.net/model/claude-fable.md) | 39.5% | 36.3% | estimated ± 6.4 pp, low confidence |
| [Claude Opus 4.5](https://benchgap.net/model/claude-opus-4-5.md) | 14.3% | 0.0% | estimated ± 6.4 pp, low confidence |
| [Claude Opus 4.6](https://benchgap.net/model/claude-opus-4-6.md) | 12.7% | 0.0% | estimated ± 6.4 pp, low confidence |
| [Claude Opus 4.7](https://benchgap.net/model/claude-opus-4-7.md) | 19.0% | 0.0% | estimated ± 6.4 pp, low confidence |
| [Claude Sonnet 4.5](https://benchgap.net/model/claude-sonnet-4-5.md) | 2.4% | 0.0% | estimated ± 6.4 pp, low confidence |
| [Gemini 3.5 Flash](https://benchgap.net/model/gemini-3-5-flash.md) | 4.8% | 0.0% | estimated ± 6.4 pp, low confidence |
| [GLM-5.2](https://benchgap.net/model/glm-5-2.md) | 9.5% | 0.0% | estimated ± 6.4 pp, low confidence |
| [GPT-5](https://benchgap.net/model/gpt-5.md) | 12.7% | 0.0% | estimated ± 6.4 pp, low confidence |
| [GPT-5.2](https://benchgap.net/model/gpt-5-2.md) | 23.0% | 0.0% | estimated ± 6.4 pp, low confidence |
| [GPT-5.4](https://benchgap.net/model/gpt-5-4.md) | 25.4% | 0.0% | estimated ± 6.4 pp, low confidence |
| [Grok 4.6](https://benchgap.net/model/grok-4-6.md) | 30.5% | 0.0% | estimated ± 6.4 pp, low confidence |
| [Kimi K2.6](https://benchgap.net/model/kimi-2-6.md) | 2.4% | 0.0% | estimated ± 6.4 pp, low confidence |
| [Muse Wonder](https://benchgap.net/model/muse-wonder.md) | 31.4% | 0.0% | estimated ± 6.4 pp, low confidence |
| [Qwen3.7 Max](https://benchgap.net/model/qwen3-7-max.md) | 9.5% | 0.0% | estimated ± 6.4 pp, low confidence |

## Fitted on

| Model | EBR-bench | ExploitBench |
|---|---|---|
| [GPT-5.6 Sol](https://benchgap.net/model/gpt-5-6-sol.md) | 44.8% | 73.5% |
| [GPT-6.1 Sol](https://benchgap.net/model/gpt-6-1-sol.md) | 54.3% | 99.7% |
| [GPT-6.1 Terra](https://benchgap.net/model/gpt-6-1-terra.md) | 78.1% | 97.1% |
| [GPT-6.2 Astra](https://benchgap.net/model/gpt-6-2-astra.md) | 92.4% | 100.0% |
| [GPT-6 Astra](https://benchgap.net/model/gpt-6-astra.md) | 76.2% | 100.0% |
