Benchmark
Sayf MMLU computer_security (5-shot) accuracy
Higher scores are better.
- Models measured
- 6
- Scores recorded
- 6
- Sources
- 1
- IQ weight
- 0.92
Score distribution
- 0.87
- Mean
- 0.88
- Median
- 0.031
- Std dev
- 0.86
- 10th percentile
- 0.9
- 90th percentile
- 0.81
- Min
This benchmark has a weight of 0.92 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 4.6mmlu-cs· Sayf cybersecurity evaluations data docs | 0.9 |
| 2 | Gemma 4 31B ITmmlu-cs· Sayf cybersecurity evaluations data docs | 0.9 |
| 3 | Qwen3.6 35B A3Bmmlu-cs· Sayf cybersecurity evaluations data docs | 0.88 |
| 4 | GPT 5.4mmlu-cs· Sayf cybersecurity evaluations data docs | 0.87 |
| 5 | GPT OSS 20bmmlu-cs· Sayf cybersecurity evaluations data docs | 0.86 |
| 6 | Llama 3.3 70B Instructmmlu-cs· Sayf cybersecurity evaluations data docs | 0.81 |
Data credits: Sayf-eval. Copyright (c) 2026 Qatar Computing Research Institute (QCRI). MIT License. (licence)