Generalization V2 (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 eight similar candidates. V2 uses 1,247 validated prompts.
Metric: Inverse-Rank Score. Source: github.com. Status: saturation imminent. 30 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.6 (Thinking, High) | 80.6 |
| 2 | GPT-5.4 (xHigh) | 80 |
| 3 | Gemini 3.1 Pro (Preview) | 79.4 |
| 4 | GPT-5.4 (High) | 77.4 |
| 5 | Claude Sonnet 4.6 (Thinking, High) | 76.3 |
| 6 | GLM-5.1 | 69.8 |
| 7 | Kimi K2.5 (Thinking) | 69.4 |
| 8 | Claude Opus 4.6 (Non-reasoning) | 68.8 |
| 9 | Qwen 3.5 397B A17B | 65.1 |
| 10 | DeepSeek V3.2 | 65 |
| 11 | Grok 4.20 0309 (Reasoning) | 63.8 |
| 12 | Gemini 3.1 Flash Lite (Preview) | 63.3 |
| 13 | GPT-5.4 Mini (xHigh) | 61.7 |
| 14 | Qwen 3.6 Plus | 59.5 |
| 15 | Seed 2.0 Pro | 57.1 |
Interactive version: theaggregate.ai/benchmark?slug=generalization-v2-lechmazur · How the rankings work · Data refreshed daily, snapshot 2026-07-22.