Which time windows can I compare?

Rankings cover the last 24 hours, 7 days and 30 days. The change column compares each model's token volume with the previous window of the same length.

Are these rankings the same as benchmark leaderboards?

No. Benchmarks measure quality on fixed tests; these rankings measure adoption. Use both: model pages show benchmark results, where available, next to pricing and providers.

Can I use the top-ranked models through one API?

Yes. Every ranked model is available through LLM Gateway's OpenAI-compatible API at https://api.llmgateway.io/v1 with a single API key.

Live from the gateway

LLM Rankings

Which models and apps developers actually use in production. Ranked by real token volume routed through LLM Gateway — not benchmarks, not vibes.

Live gateway traffic, updated every few minutes

Tokens processed

Requests routed

Models ranked

Token volume by model

Tokens processed through LLM Gateway for the top models over the selected window. Hover or focus a model to isolate its usage.

Model leaderboard

Every public model ranked by real token volume, with change vs the previous window.

Provider market share

Share of token volume routed to each provider.

Top apps

All-time token volume by app and source.

View all

1openai-sdk3.5M

How the LLM rankings work

Each model is ranked by the tokens it processed through LLM Gateway in the selected window, across every provider that serves it. Token volume shows where real workloads run: a model can top a benchmark and still see little production use.

What is counted

Tokens from requests routed through the gateway, grouped by model. The apps list ranks the tools that identify themselves by the tokens they send.

How to read the change column

The change compares a model's token volume with the previous window of the same length, so a new release climbing fast stands out even before it reaches the top.

FAQ

LLM rankings, answered

Can't find an answer? Contact us

Models are ranked by the total tokens routed through LLM Gateway in the selected window. Every request counts, whichever provider served it, so the ranking reflects what developers run in production rather than benchmark scores.