IdeaAMBIG - Readiness Assessment - Controlled Synthetic: leaderboard
IdeaAMBIG (arXiv 2609.10539v1, Table 1): the model judges whether a research-idea specification is Ready or NotReady for implementation, on controlled synthetic specifications derived from ideation-execution trajectories. Score is Macro-F1 over 100 Ready and 100 NotReady specifications, in percent; higher is better. Zero-shot fixed prompts at temperature 0.
Metric: Macro-F1 (%; 100 Ready and 100 NotReady specifications per subset). Source: arxiv.org. Saturation forecast: Around February 2027. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.6 Sol | 86.4 | #22 |
| 2 | Claude Sonnet 5 | 76.2 | #65 |
| 3 | Gemini 3.1 Pro (Preview) | 67.3 | #64 |
| 4 | DeepSeek V3.2 | 67.2 | #210 |
| 5 | GLM-5.2 | 65.9 | #80 |
| 6 | DeepSeek R1 0528 | 59.9 | #232 |
| 7 | Gemma 4 31B (IT) | 56.9 | #178 |
| 8 | Qwen 3.5 397B A17B | 55.9 | #154 |
| 9 | Qwen 3.5 9B | 54.3 | #391 |
| 10 | Qwen 3 8B | 48.7 | #678 |
| 11 | GPT-OSS-120B | 44.4 | #346 |
| 12 | Kimi K3 | 43.1 | #35 |
| 13 | Qwen 3 32B | 37 | #449 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=ideaambig-readiness-assessment-controlled-synthetic · How It Works · Data refreshed daily, snapshot 2026-11-05.