MPCI-Bench - Seed Tier Probing: leaderboard
Metric: Probe Accuracy (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5 | 97.3 |
| 2 | GPT-4o | 96 |
| 3 | Qwen 3 VL 8B | 95.3 |
| 4 | Qwen 3 VL 4B | 93.9 |
| 5 | Gemma 3 4B (IT) | 91 |
| 6 | Mistral Large 3 | 90.5 |
| 7 | Gemma 3 12B (IT) | 84.7 |
| 8 | Gemma 3 27B (IT) | 83.3 |
Interactive version: theaggregate.ai/benchmark?slug=mpci-bench-seed-tier-probing · How It Works · Data refreshed daily, snapshot 2026-09-25.