Grok 4.5
Grok 4.5 is a strong all-rounder: #37 of 406 by SIscope score. Strongest at reasoning & knowledge, weaker at coding. Mid-range on price.
Shape
Each spoke is a category, 0 at the centre, 100 at the edge.
Category scores
Percentiles: 90 means better than about 90% of tested models.
What goes into each category · Compare with DeepSeek V4.1 Flash
Human-vote score over time
Text arena, recent updates.
Coding votes over time
WebDev arena, recent updates.
Independent test runs
Run by Epoch AI on the same harness for every model. Bars show the share of questions answered correctly.
Human-vote arenas
People compare two anonymous answers and pick the better one. Scores work like chess ratings.
| Arena | Rank | Score | Range (95%) | Votes | Updated |
|---|---|---|---|---|---|
| Text arena Everyday chat, blind human votes | #58 of 419 | 1448 | 1443–1452 | 39,801 | Oct 28, 2026 |
| WebDev arena Building web apps, blind human votes | #46 of 152 | 1553 | 1546–1560 | 11,018 | Oct 28, 2026 |
| Vision arena Questions about images, blind human votes | #31 of 167 | 1287 | 1280–1294 | 12,599 | Oct 19, 2026 |
| Agent arena Real multi-step agent sessions | Not on this board yet | ||||
Source: Arena (formerly LMArena) leaderboard dataset, CC BY 4.0. Fetched Nov 4, 2026.
Price
List price from the OpenRouter models API (x-ai/grok-4.5), Nov 4, 2026. Providers can charge differently.