EN中
← All visual benchmarksArena / 2026-10-22
Arena · GENERAL

General model preference

Blind pairwise human preference, not percent accuracy.

Hover or tap a bar to inspect. Select up to 3 models to highlight and compare.
Model scores10 models
General model preference. Zoomed horizontal axis, from 1485.9 to 1530.9ZOOMED X-AXIS · 1485.9 – 1530.9 · NOT FROM ZERO1485.91497.21508.41519.61530.9Gemini 4 Argon (High)Gemini 4 Argon (High) · 1525.7 Arena Rating · zoomed horizontal axis1525.7Claude Opus 5.5 (High)Claude Opus 5.5 (High) · 1516.2 Arena Rating · zoomed horizontal axis1516.2Claude Fable 5.1 (Max)Claude Fable 5.1 (Max) · 1512.5 Arena Rating · zoomed horizontal axis1512.5Claude Opus 5 (Max)Claude Opus 5 (Max) · 1508.3 Arena Rating · zoomed horizontal axis1508.3Claude Opus 5 (High)Claude Opus 5 (High) · 1503.3 Arena Rating · zoomed horizontal axis1503.3Claude Opus 4.6 (High)Claude Opus 4.6 (High) · 1501.5 Arena Rating · zoomed horizontal axis1501.5Muse Wonder (Max)Muse Wonder (Max) · 1500.5 Arena Rating · zoomed horizontal axis1500.5Gemini 3.8 Flash (High)Gemini 3.8 Flash (High) · 1498.3 Arena Rating · zoomed horizontal axis1498.3Claude Opus 4.6Claude Opus 4.6 · 1496.8 Arena Rating · zoomed horizontal axis1496.8Muse Spark 1.3 (Max)Muse Spark 1.3 (Max) · 1491.1 Arena Rating · zoomed horizontal axis1491.1
Choose a bar for details.

How to read this chart

Official top-ranked excerpt. Human preference ratings are not accuracy percentages; IPS is a distinct metric.

Score unit: Arena Rating. Values can only be compared within the same task and protocol.

Source publication: 2026-10-22. Higher values indicate better results for the reported metric.

Source: Arena · CC BY 4.0. Redrawn from released data; reduced to selected models and rounded. Not affiliated with the source organization.

← Explore all benchmark charts