SAFARI - Task 2 Safety-Goal Agreement (CoT): leaderboard
Automotive hazard analysis and risk assessment (ISO 26262 HARA) with chain-of-thought prompting: the derived safety goal scored on Safety Goal Alignment with the reference, judged by Qwen3.6-Max calibrated against certified functional-safety engineers. Higher is better. Source: arxiv.org/abs/2609.20584.
Metric: Safety Goal Alignment (%, judge Qwen3.6-Max). Source: arxiv.org. Saturation forecast: Not forecast. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.5 122B A10B | 55.7 | #184 |
| 2 | DeepSeek V4 Flash | 49.7 | #155 |
| 3 | GLM-5.1 | 48.7 | #109 |
| 4 | DeepSeek V4 Pro | 43.8 | #112 |
| 5 | Qwen 3.5 397B A17B | 43.7 | #154 |
| 6 | GPT-5.4 Mini | 43.3 | #228 |
| 7 | Gemini 3.5 Flash | 31.1 | #66 |
| 8 | GPT-5.5 | 30.4 | #33 |
| 9 | MiniMax-M2.7 | 22.3 | #270 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=safari-task-2-safety-goal-agreement-cot · How It Works · Data refreshed daily, snapshot 2026-11-05.