OpenAI: gpt-oss-120b Performance
Try gpt-oss-120b on us. Starter credit covers your first requests, then one key reaches all 24 provider routes with automatic failover.
openai/gpt-oss-120b
Measured performance
Continuously sampled p50/p95 time-to-first-token (TTFT), effective throughput, and success rate for OpenAI: gpt-oss-120b. Effective throughput uses provider-reported output tokens over complete request time. Unsupported route and probe-configuration rows are separated from provider downtime, and no prompt or output content is stored.
| Provider | p50 TTFT | p95 TTFT | Effective throughput | Uptime | Config excluded | Availability samples |
|---|---|---|---|---|---|---|
| cerebras | 1287 ms | 2737 ms | 284 tok/s n=2 | 100.00% | — | 11 |
| fireworks | 985 ms | 985 ms | — | 100.00% | — | 1 |
| privatemode | 988 ms | 1445 ms | — | 100.00% | 9 provider_auth_config |
4 |
| akashml | 1064 ms | 2113 ms | — | 100.00% | — | 3 |
| nebius | 1164 ms | 1164 ms | — | 100.00% | — | 1 |
| nscale | 1225 ms | 1357 ms | 113 tok/s n=2 | 100.00% | — | 2 |
| sail-research | 1276 ms | 1307 ms | — | 100.00% | — | 2 |
| tinfoil | 1296 ms | 3249 ms | — | 100.00% | — | 9 |
| sambanova | 1301 ms | 5114 ms | — | 100.00% | — | 6 |
| mancer | 1320 ms | 7900 ms | — | 100.00% | — | 8 |
| parasail | 1388 ms | 1411 ms | — | 100.00% | — | 3 |
| scaleway | 1463 ms | 2160 ms | — | 100.00% | — | 4 |
| io-net | 1550 ms | 1659 ms | — | 100.00% | — | 2 |
| crusoe | 1571 ms | 1571 ms | — | 100.00% | — | 1 |
| databricks | 1664 ms | 3478 ms | — | 100.00% | — | 2 |
| regolo | 1858 ms | 2387 ms | — | 100.00% | — | 5 |
| cloudflare-workers-ai | 2171 ms | 2171 ms | — | 100.00% | — | 1 |
| amazon-bedrock | — | — | 186 tok/s n=2 | — | — | 0 |
Full provider & model leaderboard.
24 routes.
More routes give the auto router more room to fail over around provider 429 and 5xx responses.
Gateway overhead is measured separately.
Public status separates TLS/health overhead from full model latency so slow LLMs do not inflate the router metric.
Metadata rollups.
Status samples store latency, outcome, provider, model, route, cost, and region metadata only.
View public status or inspect provider routes.