Model Rankings

What people actually run on api.airforce. Every number here is measured from our own traffic — nothing is imported, estimated or ranked by opinion.

Top models

Token usage across all models on api.airforce.

Loading…

Leaderboard

The full table, with prompt, completion, reasoning and cached tokens side by side. Users/day is the highest number of distinct users on any single day in the window.

Loading…

Top models by task

Each task's leading models, ranked by share of spend.

Loading…

Cost per session

What one session typically costs, by session length.

Loading…

Market share

Share of requests by model maker over time.

Loading…

Benchmarks

Independent benchmark scores next to measured usage.

Loading…

Fastest models

Median time to first token and sustained throughput, measured on delivered responses only.

Loading…

Languages

Model usage by natural language.

We do not detect the language of prompts. Doing so would mean inspecting message content, which we deliberately do not store — so this section stays empty rather than showing a guess.

Programming

Model usage by programming language.

We do not detect programming languages in prompts, for the same reason as above: it would require inspecting message content, which we do not store.

Context length

Requests by prompt length.

Loading…

Tool calls

Tool calls across models.

Loading…

Images

Images generated across models.

Loading…

Top apps

Applications sending the most traffic.

Loading…