HorizonBench: leaderboard
HorizonBench (2026) tests long-horizon personalization: models read six months of synthetic user conversations in which life events change preferences, then pick the response that fits the current preference among five options. Accuracy in percent over all items; higher is better.
Metric: Accuracy (%) on all 4,245 items (360 users, three generator models) of HorizonBench (5-option questions on which assistant response fits the user's current preference, given six months of synthetic conversation history, about 163K tokens and 4,300 turns; pre-evolution values as hard distractors; items kept only if five validation models all failed them without the history; chance 20); higher is better. Source: arxiv.org. Saturation forecast: Not forecast. 21 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.5 | 52.8 | #160 |
| 2 | Claude 3.7 Sonnet | 47.8 | #470 |
| 3 | Claude Opus 4 | 45.1 | #326 |
| 4 | Gemini 3.1 Pro (Preview) | 41.2 | #99 |
| 5 | Gemini 3 Pro | 40.9 | #145 |
| 6 | Claude Sonnet 4.5 | 34 | #261 |
| 7 | Claude 3.5 Haiku | 32.5 | #1580 |
| 8 | Claude Sonnet 4 | 32.2 | #410 |
| 9 | Claude 3.5 Sonnet (20240620) | 29.8 | #692 |
| 10 | Gemini 2.0 Flash | 24.7 | #723 |
| 11 | GPT-4o | 24.6 | #751 |
| 12 | Gemini 2.5 Pro | 22.8 | #276 |
| 13 | Gemini 2.0 Flash Lite | 20.9 | #1050 |
| 14 | Gemini 2.5 Flash Lite | 19.5 | #964 |
| 15 | O3 | 16.5 | #237 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=horizonbench · How It Works · Data refreshed daily, snapshot 2026-11-21.