ASCIIBench: leaderboard
A/B voting platform for AI-generated ASCII art. Models compete head-to-head and are ranked by ELO rating. Tests creative visual generation constrained to text characters.
Metric: ELO Rating. Source: www.asciibench.com. Status: saturation imminent. 23 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.5 | 1702 |
| 2 | GPT-5.5 | 1682 |
| 3 | Claude Opus 4.7 | 1674 |
| 4 | Gemini 3 Pro (Preview) | 1674 |
| 5 | Claude Opus 4.1 | 1638 |
| 6 | Gemini 3 Flash (Preview) | 1617 |
| 7 | Claude Sonnet 4.5 | 1599 |
| 8 | Gemini 2.5 Pro | 1583 |
| 9 | GPT-5.2 | 1517 |
| 10 | GPT-5 Codex | 1479 |
| 11 | GPT-4o ChatGPT | 1472 |
| 12 | GLM-4.7 | 1408 |
| 13 | Kimi K2.5 | 1352 |
| 14 | MiniMax-M2.1 | 1335 |
| 15 | Grok 4 | 1328 |
Interactive version: theaggregate.ai/benchmark?slug=asciibench · How It Works · Data refreshed daily, snapshot 2026-09-05.