InferWeave

Capacity

Ready capacity, queued demand and recent movement for each model and card — the detail view behind the dashboard's headline number.

Fixture data (no live surface yet).

Per-resource capacity

  • demo/7b on nvidia:h100
    • Ready now: 12
    • Queued demand: 4
    • ready: up across the recent window
    • queued: up across the recent window
    • 3 samples, 2023-11-14T22:12:15.000Z to 2023-11-14T22:13:15.000Z
  • demo/7b on amd:mi300
    • Ready now: 3
    • Queued demand: 9
    • ready: down across the recent window
    • queued: up across the recent window
    • 2 samples, 2023-11-14T22:12:45.000Z to 2023-11-14T22:13:15.000Z
  • demo/3b on nvidia:h100
    • Ready now: 6
    • Queued demand: 0
    • ready: flat across the recent window
    • queued: down across the recent window
    • 2 samples, 2023-11-14T22:12:45.000Z to 2023-11-14T22:13:15.000Z
  • demo/1b on cpu:64c
    • Ready now: 1
    • Queued demand: 0
    • ready: no movement yet (single sample)
    • queued: no movement yet (single sample)
    • single sample at 2023-11-14T22:13:15.000Z