aaplot

Unofficial terminal dashboard for Artificial Analysis model data — cost vs intelligence quadrant charts, rankings, side-by-side compares, and media arena ELOs. Not affiliated with Artificial Analysis.

Requires Node.js 20+.

Install

npm i -g aaplot
aaplot

Or run without installing:

npx aaplot

Preview without an API key:

aaplot --demo

API key

Every live request needs a free key from artificialanalysis.ai/data-api (100 requests/day, shared per organisation).

Set AA_API_KEY, or paste the key on first run. aaplot stores it as plaintext JSON in the user config directory (env-paths name aaplot): %APPDATA%\aaplot-nodejs\Config on Windows, ~/.config/aaplot-nodejs on Linux, ~/Library/Preferences/aaplot-nodejs on macOS. POSIX files are mode 0600. On Windows there is no extra ACL step — the file inherits your user-profile permissions. Keys are free and revocable.

Usage

aaplot                 # TUI
aaplot --demo          # synthetic data, no key, no quota
aaplot --ascii         # ASCII plot markers (no braille)
aaplot --offline       # cached data only
aaplot --creator openai --min-quality 40 --cheap
aaplot --help

In the TUI: 1–4 jump Models / Plot / Compare / Media, ? keybindings, r refresh (quota-aware; bypasses a fresh cache), q or ctrl+c quit.

Pin models on Models (p / space) to drive both the Plot pin set and the Compare table. Compare marks the winner of each metric with ★.

Braille plots need a VT-capable terminal. Windows Terminal and VS Code use braille automatically. Other Windows consoles and TERM=dumb fall back to ASCII; --ascii forces that everywhere.

Data and attribution

Data: Artificial Analysis. Licensed under their API terms. Footer on every screen credits them.

--demo and tests use synthetic fixtures only. aaplot does not scrape the website or ship Artificial Analysis datasets.

License

MIT