JobCannon Protocol Sensitivity - Multiple Intelligences Options Reversed: leaderboard
A model answers the Multiple Intelligences questionnaire one item per request, then again with the answer options listed in reverse order; the number is the share of items whose chosen answer changes. It isolates option-order sensitivity, where the family headline pools every protocol change.
Metric: Share of Multiple Intelligences items whose answer changes when the answer options are listed in reverse order (0-1, lower is better; table F13). Source: huggingface.co. 16 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 5.5 | 0.05 | #5 |
| 2 | Kimi K2.5 | 0.1 | #256 |
| 3 | Gemini 2.5 Pro | 0.13 | #266 |
| 4 | GPT-6 Astra | 0.15 | #6 |
| 5 | Claude Fable 5.1 | 0.15 | #15 |
| 6 | Gemma 3 27B (IT) | 0.2 | #1289 |
| 7 | Mistral Large 3 | 0.2 | #817 |
| 8 | Qwen 3 VL 235B A22B | 0.28 | #479 |
| 9 | Nova Pro | 0.28 | #1555 |
| 10 | GPT-OSS-120B | 0.35 | #700 |
| 11 | Claude Haiku 4.5 | 0.38 | #568 |
| 12 | GLM-5 | 0.38 | #260 |
| 13 | DeepSeek R1 | 0.45 | #473 |
| 14 | Llama 4 Maverick | 0.48 | #1099 |
| 15 | Qwen 3 32B | 0.6 | #960 |
Interactive version: theaggregate.ai/benchmark?slug=jobcannon-protocol-sensitivity-multiple-intelligences-options-reversed · How It Works · Data refreshed daily, snapshot 2026-11-13.