Benchmarks
347 sources · 1,258 models × 3,038 benchmarks · 158,912 scoresThe explicit version of Mercor's 12 enterprise-risk tasks, in which the risk is stated in one sentence of the user's request. The score is the share of 36 runs (three per task) in which the agent saved the harmful change; lower is better. Same harness, reasoning settings and judging as the implicit version. Diagnostic.
No signal or difficulty estimate. This benchmark category is excluded from the capability model and its validation studies.
| # | Model | Setting | Score | Source |
|---|---|---|---|---|
| 1 | Claude Fable 5.1 | max · Mercor enterprise-risk harness; fresh copy of a simulated law, banking or consulting workspace per run and the same tools for every model; one-line system prompt | 0.0 | |
| 2 | GPT-6.1 Sol | max · Mercor enterprise-risk harness; fresh copy of a simulated law, banking or consulting workspace per run and the same tools for every model; one-line system prompt | 0.0 | |
| 3 | GPT-6 Astra | max · Mercor enterprise-risk harness; fresh copy of a simulated law, banking or consulting workspace per run and the same tools for every model; one-line system prompt | 0.0 | |
| 4 | Claude Opus 5.5 | max · Mercor enterprise-risk harness; fresh copy of a simulated law, banking or consulting workspace per run and the same tools for every model; one-line system prompt | 5.6 |