Provider reliability · tracked daily
OpenAI: gpt-oss-120b: which provider is most reliable?
Across 18 providers, Cerebras is the most reliable way to run OpenAI: gpt-oss-120b, at 100.0% uptime and 215ms median latency. Measured directly from OpenRouter's endpoint data, tracked daily.
- Cerebras - 100.0% uptime · 215ms
- BaseTen - 100.0% uptime · 308ms
- AkashML - 100.0% uptime · 790ms
- DigitalOcean - 100.0% uptime · 963ms
- Mara - 100.0% uptime · 965ms
- Novita - 100.0% uptime · 580ms
- Groq - 100.0% uptime · 211ms
- DeepInfra - 99.9% uptime · 483ms
- Mancer 2 - 99.8% uptime · 932ms
- Parasail - 96.9% uptime · 402ms
Uptime is OpenRouter's own trailing-30-minute metric, captured once at our last daily sync - a snapshot of that window, not a live/real-time feed.
| # | Provider | Uptime (30m) | Latency p50 | Throughput |
|---|---|---|---|---|
| 01 | Cerebras | 100.0% | 215ms | 777 tok/s |
| 02 | BaseTen | 100.0% | 308ms | 189 tok/s |
| 03 | AkashML | 100.0% | 790ms | 57 tok/s |
| 04 | DigitalOcean | 100.0% | 963ms | 34 tok/s |
| 05 | Mara | 100.0% | 965ms | 246 tok/s |
| 06 | Novita | 100.0% | 580ms | 78 tok/s |
| 07 | Groq | 100.0% | 211ms | 367 tok/s |
| 08 | DeepInfra | 99.9% | 483ms | 41 tok/s |
| 09 | Mancer 2 | 99.8% | 932ms | 41 tok/s |
| 10 | Parasail | 96.9% | 402ms | 99 tok/s |
| 11 | SambaNova | 95.6% | 988ms | 359 tok/s |
| 12 | Nebius | 95.3% | 483ms | 151 tok/s |
| 13 | Together | 94.8% | 550ms | 125 tok/s |
| 14 | CoreWeave | 94.5% | 460ms | 30 tok/s |
| 15 | SiliconFlow | 92.5% | 2466ms | 21 tok/s |
| 16 | Phala | 89.9% | 1053ms | 88 tok/s |
| 17 | Amazon Bedrock | 86.7% | 825ms | 150 tok/s |
| 18 | 86.0% | 348ms | 99 tok/s |
Uptime over time
| # | Model | Uptime (%) |
|---|---|---|
| 01 | AkashML | 100% |
| 02 | BaseTen | 100% |
| 03 | Cerebras | 100% |
| 04 | DeepInfra | 100% |
| 05 | CoreWeave | 94% |
| 06 | Amazon Bedrock | 87% |
Daily tracking began Aug 9- this becomes a live chart once there's a third day to plot.
Frequently asked questions
Which provider is most reliable for OpenAI: gpt-oss-120b?
Cerebras is the most reliable provider for OpenAI: gpt-oss-120b, at 100.0% uptime (OpenRouter's trailing-30-minute metric, snapshotted at our last daily sync) and 215ms median latency, out of 18 providers tracked.
What is OpenAI: gpt-oss-120b's uptime?
The best-performing provider for OpenAI: gpt-oss-120b reports 100.0% uptime, tracked daily.