Weights & Biases Uptime History
Measured by PulsAPI's own checks, not reported by Weights & Biases. Every figure below is computed over the 135 days in the last 365 that PulsAPI actually observed, from 2026-05-30 to 2026-10-11. Days nobody watched are left out rather than counted as perfect.
Weights & Biases Uptime at a Glance
- Measured uptime
- 100.00%
- Days with an issue
- 14
- Longest outage
- 5d 22h
- Mean recovery
- 10h 26m
across 135 observed days
days were not fully operational
single longest resolved incident
average time to resolve
Monthly Weights & Biases Uptime
Each month is the mean of its observed days. Months with few observed days are marked, because a short month is not evidence of a good one.
| Month | Uptime | Uptime bar | Days observed |
|---|---|---|---|
| October 2026 | 100.00% | 11partial | |
| September 2026 | 100.00% | 30 | |
| August 2026 | 100.00% | 31 | |
| July 2026 | 100.00% | 31 | |
| June 2026 | 100.00% | 30 | |
| May 2026 | 100.00% | 2partial |
Get an email when Weights & Biases moves
Every figure on this page came from a check we ran. Those same checks can reach your inbox the moment Weights & Biases stops being operational, so the next dip arrives as a notification rather than a row you find later.
Free, no password, no card. Unsubscribe in one click.
How This Weights & Biases Uptime Figure Is Measured
PulsAPI checks Weights & Biases continuously (every 60 seconds while it is moving) and records the result. A day's uptime is the share of that day's checks that came back fully operational, so a 20-minute outage shows up as a dip rather than being rounded away. The window figure is the mean of those daily values.
Two things this deliberately is not. It is not Weights & Biases's own published uptime, because vendors generally measure against their own SLA definitions, which usually exclude scheduled maintenance and often exclude partial degradation. And it is not a promise about the future: 135 days of history describes what happened, and the useful way to read it is alongside the incidents behind the dips.
Uptime measured this way is a mean of component uptimes where a vendor publishes components, so coverage matters as much as the percentage. That is why the observed-day count sits beside every figure on this page rather than in a footnote.
These measurements are published under CC BY 4.0. Quote or republish them freely, with attribution to PulsAPI.