Google's gemini-3-pro leads the Vision Arena at 1,288; Gemini models hold four of the top six places

What the chart shows

The Vision Arena applies the Arena’s blind, vote-based ranking to conversations that include images. In February 2026 gemini-3-pro led with 1,288, followed by gemini-3.1-pro-preview (1,278), gemini-3-flash (1,276) and gpt-5.2-chat-latest-20260210 (1,271). Gemini models hold four of the top six places.

The top 15 span 55 points, down to gpt-5.2 at 1,233. Below the top four, eleven models sit between 1,233 and 1,260, and many of their 95% confidence intervals overlap.

Key numbers

How to read it

• Top 15 models on the Arena vision leaderboard (arena.ai) as of February 2026.
• The report describes this leaderboard as ranking image generation models. The Vision Arena compares answers to prompts that contain images, and the models listed are multimodal chat models, so we describe it that way.
• Elo-like ratings have no true zero, so the chart shows dots on an axis starting at 1,220: compare the differences between models.