CESBench - Multiple Choice: leaderboard
Metric: Accuracy (%; the 209 four-option multiple-choice items whose distractors encode practitioner misconceptions; 380 expert-written items on cryptographic engineering security for IoT devices; zero-shot, temperature 0, no tools, provider default reasoning mode; judgment and scenario answers scored by a Qwen3.5-397B judge from outside the evaluated set). Source: arxiv.org. Saturation forecast: Estimated already saturated. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GLM-5.2 | 98.6 |
| 2 | Gemini 3.7 Flash | 98.6 |
| 3 | GPT-5.6 Luna | 97.6 |
| 4 | DeepSeek V4 Pro | 97.1 |
| 5 | Kimi K2.6 | 97.1 |
| 6 | MiniMax-M2.5 | 95.2 |
| 7 | Llama 4 Maverick | 93.3 |
| 8 | DeepSeek V3 | 91.9 |
| 9 | GLM-4 32B (0414) | 88.5 |
| 10 | Hunyuan A13B-Instruct | 85.6 |
| 11 | Ling-flash-2.0 | 83.3 |
Interactive version: theaggregate.ai/benchmark?slug=cesbench-multiple-choice · How It Works · Data refreshed daily, snapshot 2026-09-26.