About
What is Avenvote?
Compare the models developers actually use most, the ones people rate best in blind tests, and the ones that give the most for the price, side by side. We cover language models as well as image and video generation models.
How we rank
- Usage — Tokens developers worldwide actually used through OpenRouter in the last 7 days. For images and video, each model's share of generations on Vercel AI Gateway.
- Performance and quality — Arena scores from people who compare two outputs for the same prompt without knowing which model made them, then vote.
- Value — Models from the performance ranking, re-sorted by output price, cheapest first.
- Prices — Per-1M-token prices from the OpenRouter model list, converted at fixed exchange rates.
Rankings refresh every day at 00:00 UTC.
Rankings are not for sale
Rankings are calculated from the data above only. No company can pay to move a model up or down.
Some official-site links are affiliate links. If you sign up or pay through one, Avenvote may earn a commission. Commissions never affect rankings or their order.
Data sources
- Usage: OpenRouter Data API (openrouter.ai/rankings), CC BY 4.0
- Image and video usage share: © 2026 Vercel, "AI Gateway Leaderboard Data", CC BY 4.0
- Performance and quality: Arena (lmarena-ai/leaderboard-dataset), CC BY 4.0
- Prices: OpenRouter model list
- Logos: LobeHub Icons (MIT) and company websites. Model names and logos are trademarks of their companies. Avenvote is not affiliated with them.
In 9 languages
Available in English, 简体中文, 繁體中文, 日本語, 한국어, Deutsch, Français, Español and Português. One-line reviews are translated into the other languages as soon as they’re posted.
Who runs it
Avenvote is run independently by the Avenvote team. Found a data error or have a suggestion? Contact us.