ReCon: leaderboard

Metric: Balanced accuracy (%) of CONFORMITY or NON-CONFORMITY verdicts against certified-auditor assessments for the 93 ISO/IEC 27002:2022 controls on three real policy sets, over the study retrieval configurations (overall figure of Table 5; 50% is chance); retrieval-augmented prompts, local CPU inference at temperature 0; higher is better. Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 8 models tracked.

Top models

#ModelScore
1ReCon granite4.1 8B (Ollama build)75
2ReCon granite4.1 3B (Ollama build)73.4
3ReCon granite3.3 8B (Ollama build)68
4ReCon llama3.2 3B (Ollama build)64.4
5ReCon phi4 14B (Ollama build)59.9
6ReCon mistral 7B (Ollama build)54.6
7ReCon gemma3 4B (Ollama build)54.3
8ReCon gemma3n E2B (Ollama build)52.4

Interactive version: theaggregate.ai/benchmark?slug=recon · How It Works · Data refreshed daily, snapshot 2026-09-29.