The board
opt-inEveryone who ran tokenchit publish, or a bare tokenchit generate, which ends by calling it. A usage count over the selected window, not a skill score.
Get your own card →filterslast 30d · all agents
GitHub sign-in proves the handleself-reported, handle unprovedNeither audits the figures — see the notes below the table.
| rank | developer | agent mix | last 30d | tokens | equiv. cost | streak |
|---|---|---|---|---|---|---|
| 1– | 38.9B | $35,639 | 0d | |||
| 2– | 32.4B | $7,194 | 0d | |||
| 3– | 24.4B | $15,477 | 11d | |||
| 4– | 12.6B | $4,813 | 0d | |||
| 5▲1 | 3.30B | $980 | 0d | |||
| 6▼1 | 3.27B | $98 | 0d | |||
| 7▲1 | 1.55B | $573 | 0d | |||
| 8▲2 | 727M | $234 | 0d | |||
| 9▲2 | 137M | $116 | 0d | |||
| 10▲2 | 33.4M | $1 | 1d |
- claude-code
- codex
How this is ranked
Verified rows first, then tokens over the selected window. Signing in is the only thing that ties a row to a GitHub account, so it is the only thing that can carry a position — an unverified row still appears with its figures, it just cannot outrank a verified one. Every column covers the same window as the rank.
A cli badge means the numbers were self-reported without a GitHub sign-in; those rows stay on the board rather than being hidden, because a visible unverified row is more honest than a quietly filtered one. Streak is the current run of active days, which is not a windowed figure. Submissions far outside the range of real usage are held for review and do not appear until a person has looked. Figures are self-reported and bounded for plausibility, not audited.