Benchmarks · agentic
CheatBench
Uses the published cheating-attempt rate; lower is better. Score: cheating rate. Collected by Jake Boggs' benchmark dashboard.
9 models · 9 results · data as of 2026-10-29 · cheatbench.ai ↗
Leaderboard
Gemini 4 Argon max
All results
9 rows| # | Model | Effort | CheatBench |
|---|---|---|---|
| 1 | Gemini 4 Argon | max | 8.6% |
| 2 | Opus 5.5 | max | 11.2% |
| 3 | Muse Wonder (Max) | max | 12.7% |
| 4 | Fable 5.1 | max | 45.8% |
| 5 | GPT-6 Astra | max | 47.4% |
| 6 | GPT-6.1 Sol | max | 49.3% |
| 7 | Kimi K3 | max | 70.0% |
| 8 | DeepSeek V4 Pro | max | 75.1% |
| 9 | Grok 4.7 | max | 78.0% |