Chatbot Arena (Text - Hard Prompts): leaderboard
The arena.ai text board restricted to prompts that a language model classifier judges demanding against a checklist covering specificity, domain knowledge, layered reasoning, problem solving, creativity, technical accuracy and real world use. Voters pick between two anonymous replies. The Bradley-Terry rating is refit over this subset alone, and higher is better.
Metric: Arena Score. Source: arena.ai. 414 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 4 Argon (High) | 1551 | #8 (Gemini 4 Argon) |
| 2 | Claude Opus 4.6 (High) | 1533 | #63 (Claude Opus 4.6) |
| 3 | Claude Opus 5.5 (High) | 1533 | #3 (Claude Opus 5.5) |
| 4 | Claude Fable 5 (High) | 1530 | #16 (Claude Fable 5) |
| 5 | Claude Opus 4.6 | 1527 | #63 |
| 6 | Claude Opus 4.7 (High) | 1525 | #49 (Claude Opus 4.7) |
| 7 | Claude Fable 5.1 (Max) | 1520 | #9 (Claude Fable 5.1) |
| 8 | Claude Opus 4.7 | 1519 | #49 |
| 9 | Kimi K3 (Max) | 1517 | #30 (Kimi K3) |
| 10 | Gemini 3.8 Flash (High) | 1517 | #21 (Gemini 3.8 Flash) |
| 11 | Muse Spark 1.3 (Max) | 1517 | #24 (Muse Spark 1.3) |
| 12 | Claude Opus 5 (High) | 1515 | #12 (Claude Opus 5) |
| 13 | Claude Opus 5 (Max) | 1514 | #12 (Claude Opus 5) |
| 14 | Claude Opus 4.8 (High) | 1513 | #37 (Claude Opus 4.8) |
| 15 | GPT-5.6 Sol (xHigh) | 1512 | #19 (GPT-5.6 Sol) |
Interactive version: theaggregate.ai/benchmark?slug=chatbot-arena-text-hard-prompts · How It Works · Data refreshed daily, snapshot 2026-10-20.