# fireworks vs geometric — AI API latency and uptime comparison
Updated November 22, 2026. As of November 22, 2026, fireworks has lower edge latency (time-to-first-byte) than geometric in 3 of 4 measured region(s). From Asia (Tokyo), fireworks responds in 16 ms p50 vs 479 ms.

| Region | fireworks p50 | geometric p50 | Faster |
| --- | --- | --- | --- |
| Asia (Tokyo) | 16 ms | 479 ms | fireworks |
| Europe (Germany) | 90 ms | 10 ms | geometric |
| South America (Santiago) | 262 ms | 414 ms | fireworks |
| US (Central) | 30 ms | 206 ms | fireworks |


## Does fireworks or geometric have better uptime?

Over the last 24 hours fireworks and geometric are level on availability at 100% of probes succeeding across 4 region(s). Neither is separated from the other by uptime at this sample size.

| Provider | Probes succeeding | Worst region | there | Samples |
| --- | --- | --- | --- | --- |
| fireworks | 100% | Asia (Tokyo) | 100% | 1152 |
| geometric | 100% | Asia (Tokyo) | 100% | 1152 |


Observed availability over the last 24 hours, not a contractual SLA. Longer windows: https://llmlatency.dev/best-uptime

Time to first token (TTFT) for both, with the model named: https://llmlatency.dev/time-to-first-token
Source: https://llmlatency.dev/compare/fireworks-vs-geometric