Benchmark
Sayf CTI-Bench RCM (CWE mapping) accuracy
Higher scores are better.
Benchmark sourceScores from Sayf cybersecurity evaluations data docs.
- Models measured
- 6
- Scores recorded
- 6
- Sources
- 1
- IQ weight
- 0.91
Score distribution
- 0.701
- Mean
- 0.718
- Median
- 0.044
- Std dev
- 0.659
- 10th percentile
- 0.754
- 90th percentile
- 0.631
- Min
This benchmark has a weight of 0.91 for chat models, measured over 6 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | Score |
|---|---|---|
| 1 | Claude Sonnet 4.6rcm· Sayf cybersecurity evaluations data docs | 0.754 |
| 2 | GPT 5.4rcm· Sayf cybersecurity evaluations data docs | 0.742 |
| 3 | Qwen3.6 35B A3Brcm· Sayf cybersecurity evaluations data docs | 0.718 |
| 4 | Gemma 4 31B ITrcm· Sayf cybersecurity evaluations data docs | 0.705 |
| 5 | GPT OSS 20brcm· Sayf cybersecurity evaluations data docs | 0.659 |
| 6 | Llama 3.3 70B Instructrcm· Sayf cybersecurity evaluations data docs | 0.631 |
Data credits: Sayf-eval. Copyright (c) 2026 Qatar Computing Research Institute (QCRI). MIT License. (licence)