ComfyUI usage rankings put Google's Nano Banana family on top, ahead of arena picks
cremebait15 · reddit · 2026-11-22
ComfyUI published official model usage rankings (router.comfy.org), and they diverge sharply from arena-style leaderboards:
- Nano Banana Pro (Gemini 3 Pro Image) leads with 19% of all image generations; 36% of image users ran it at least once
- Nano Banana 2.1 (12%) and Nano Banana 2 (Gemini 3.1 Flash Image, 12%) take second and third — Google models account for nearly half of total generations
- Seedream 5.0 Pro follows at 14% share / 20% user coverage (the post's "Seedream 2.5" is a typo; the chart shows 5.0 Pro)
- OpenAI's GPT Image 2.5 Sunburst/Flare plus GPT Image 2 total roughly 21%
Notably, Arena ranked Omni first and Artificial Analysis ranked Wan 3 first, but real-world usage data shows the Nano Banana family is the dominant workhorse — a striking reminder that preference-based leaderboards and actual usage tell different stories.
More from Models
- Fine-tune gotcha: HF export silently drops Qwen3's MTP head, slowing decode — AIFlow_ML · 2026-11-22
- Designer KOL: Google's model pace and quality suggest it will soon 'mog them all' — AIandDesign · 2026-11-22
- Models play a literary 'listening at doors' game — Opus 4.5 gets recognized by its timbre — repligate · 2026-11-22
- Asking Claude to 'understand the entire mathematical platonic realm, make no mistakes' — burny_tech · 2026-11-22
- Twitter talks about Google Neon, but Google AI's landing page pushes Carbon — annetgriffin · 2026-11-22
- Viral claim: Claude produced a 250,000-page Riemann hypothesis proof — tinyfool · 2026-11-22