Generalization V1 (Lechmazur): leaderboard
Tests whether LLMs can infer a specific latent theme from limited examples, reject broader but incorrect patterns using anti-examples, and identify the single correct match among similar candidates. V1 uses 810 prompts.
Metric: Avg Rank (lower is better). Source: github.com. Saturation forecast: Estimated already saturated. 82 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4 (Thinking 16K) | 1.69 | #163 (Claude Opus 4) |
| 2 | Claude Opus 4 (Non-reasoning) | 1.7 | #163 (Claude Opus 4) |
| 3 | Gemini 2.5 Pro | 1.74 | #152 |
| 4 | DeepSeek R1 0528 | 1.74 | #222 |
| 5 | Gemini 2.5 Pro (Preview 05-06) | 1.75 | #147 |
| 6 | DeepSeek V3.1 (Non-reasoning) | 1.75 | #264 (DeepSeek V3.1) |
| 7 | GPT-5 (Medium) | 1.79 | #98 (GPT-5) |
| 8 | Grok 4 Fast (Reasoning) | 1.79 | #254 (Grok 4 Fast) |
| 9 | DeepSeek R1 | 1.8 | #251 |
| 10 | O1 | 1.8 | #183 |
| 11 | GLM-4.5 | 1.8 | #275 |
| 12 | O4 Mini (Medium) | 1.8 | #180 (O4 Mini) |
| 13 | O4 Mini (High) | 1.82 | #180 (O4 Mini) |
| 14 | O3 (High) | 1.82 | #124 (O3) |
| 15 | O3 (Medium) | 1.83 | #124 (O3) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=generalization-v1-lechmazur · How It Works · Data refreshed daily, snapshot 2026-10-20.