AthenaBench - CKT: leaderboard
3,000 multiple-choice cyber-threat-intelligence knowledge questions; accuracy.
Metric: Accuracy (%) on the CTI knowledge test (3,000 questions). Source: raw.githubusercontent.com. Saturation forecast: Estimated already saturated. 8 models tracked. Updated 17 Nov 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | Gemini 2.5 Flash | 85.1 | #467 | |
| 2 | Llama 3.3 70B Instruct | 81.4 | #1387 | |
| 3 | Llama 3 70B Instruct | 78.9 | #1689 | |
| 4 | Qwen 3 14B | Reasoning | 78.6 | #1264 |
| 5 | Llama-Primus-Merged | 76.3 | #2461 | |
| 6 | Qwen 3 8B | Thinking | 75.7 | #1868 |
| 7 | Qwen 3 4B 2507 Instruct | 74.7 | #2215 | |
| 8 | Llama 3.1 8B Instruct | 71.8 | #4576 |
No result here: #1 Claude Fable 5.2, #2 GPT-6.2 Astra, #3 Gemini 4 Carbon.
Interactive version: theaggregate.ai/benchmark?slug=athenabench-ckt · How It Works · Data refreshed daily, snapshot 2026-11-17.