ARCBench S6 - Full Context: leaderboard
ARCBench pi agent harness; same exact endpoint is agent and compactor, 17 task cases with only configurations having all valid repetitions included. Violation@k is percent of cases with at least one violation in k runs; lower is better. GPT-Luna retains the recorded mutable ~openai/gpt-luna-latest endpoint without an inferred dated identity. S6; k=6, GLM only, channels c1/c3; attack conditions retained as separate run configurations. Full Context.
Metric: Violation@6 (%). Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 7 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | glm-5.3-flash (ARCBench Run: endpoint=z-ai%2Fglm-5.3-flash; harness=pi; strategy=Hermes; attack=None; context=Full%20Context) | 11.8 | |
| 2 | glm-5.3-flash (ARCBench Run: endpoint=z-ai%2Fglm-5.3-flash; harness=pi; strategy=Hermes; attack=Escape; context=Full%20Context) | 11.8 | |
| 3 | glm-5.3-flash (ARCBench Run: endpoint=z-ai%2Fglm-5.3-flash; harness=pi; strategy=Hermes; attack=Legacy; context=Full%20Context) | 17.6 | |
| 4 | glm-5.3-flash (ARCBench Run: endpoint=z-ai%2Fglm-5.3-flash; harness=pi; strategy=Hermes; attack=Ignore; context=Full%20Context) | 17.6 | |
| 5 | glm-5.3-flash (ARCBench Run: endpoint=z-ai%2Fglm-5.3-flash; harness=pi; strategy=Hermes; attack=Naive; context=Full%20Context) | 23.5 | |
| 6 | glm-5.3-flash (ARCBench Run: endpoint=z-ai%2Fglm-5.3-flash; harness=pi; strategy=Hermes; attack=Fake%20completion; context=Full%20Context) | 23.5 | |
| 7 | glm-5.3-flash (ARCBench Run: endpoint=z-ai%2Fglm-5.3-flash; harness=pi; strategy=Hermes; attack=Combined; context=Full%20Context) | 23.5 |
Interactive version: theaggregate.ai/benchmark?slug=arcbench-s6-full-context · How It Works · Data refreshed daily, snapshot 2026-11-05.