CESBench - Implementation: leaderboard

The implementation sub-domain of CESBench (Zhou et al., 2026): 83 items on public-key, constant-time, post-quantum, AES and hardware implementation, across multiple-choice, judgment, scenario and code tasks. Score in percent, the mean of the four task-type scores within the sub-domain; higher is better.

Metric: Composite score (%; mean over the four task types of the item-score means within the D3 implementation sub-domain (83 items); 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: Around February 2027. 11 models tracked.

Top models

#ModelScoreOverall rank
1GLM-5.285.4#79
2DeepSeek V4 Pro78.5#112
3Gemini 3.7 Flash78.5#24
4GPT-5.6 Luna77.9#78
5Kimi K2.675.6#113
6MiniMax-M2.569.9#316
7DeepSeek V362.5#333
8Ling-flash-2.061.4#556
9GLM-4 32B (0414)57.9#540
10Hunyuan A13B-Instruct54.8#681
11Llama 4 Maverick49.6#487

No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #6 GPT-6.1 Sol.

Interactive version: theaggregate.ai/benchmark?slug=cesbench-implementation · How It Works · Data refreshed daily, snapshot 2026-11-06.