Live view of the compute pool. Updated every request; nothing here names a person or a machine.
| # | Contributor | Points | Share | Clips | Generations | LLM requests | Hosting | Nodes online | Uptime |
|---|---|---|---|---|---|---|---|---|---|
| 🥇 | contributor-268bcc |
1,293.0 | 95.7% | 92 | 175 | 2,331 · 42.5M tok | 745.4 h | 4 / 3 GPU | 3.5 d |
| 🥈 | contributor-7c8a2f |
57.6 | 4.3% | 327 | — | — | — | 1 / 1 GPU | 5.0 d |
How points work. Points are minutes of work delivered to customers: one point per minute of completed task time (clips, generations, articles) and one per minute spent answering LLM requests. Keeping a model resident is availability rather than work, so it earns one point per hour — it counts, but it cannot outrank a node that is actually running jobs.
Contributors appear under a stable pseudonym, never a name, email or machine name. If you run a node, your handle is shown on your dashboard so you can find yourself here.
Generated Mon, 12 Oct 2026 18:22:27 GMT. “Average load” is GPU utilisation as reported by each node; a resident model holds memory without necessarily loading the GPU, which is why memory-in-use and load can differ. Rented burst capacity and retired machines are not counted as nodes.