SPACE Personality Study - Base - ARC-Easy: leaderboard
SPACE v1 Table7 unmodified Instruct Base, ARC-Easy2000questions,greedy,unparseablewrong. ARC includes87unrecognizedreferencelabels countedwrong byA-Dparser. DPO/promptvariants notmerged. MixedGSM8Kextractionprotocol held.
Metric: Score (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 4 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 2.5 7B Instruct | 91.35 | #2583 |
| 2 | Qwen 2.5 3B Instruct | 89 | #5206 |
| 3 | Llama 3.1 8B Instruct | 87.6 | #4277 |
| 4 | Llama 3.2 3B Instruct | 82.3 | #7185 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=space-personality-study-base-arc-easy · How It Works · Data refreshed daily, snapshot 2026-11-12.