Start typing to search.
Ranks providers by how quickly the first generated token appears. Providers without enough recent data remain visible after ranked rows.
Groq currently ranks first at 0.14 s via groq.
| Rank | Provider / route | Time to first token | Speed measurements | Endpoint details |
|---|---|---|---|---|
| #1 | Provider / route Groq groq | Time to first token0.14 s | Speed measurements
| Endpoint details
|
| #2 | Provider / route Cerebras cerebras/fp16 | Time to first token0.23 s | Speed measurements
| Endpoint details
|
| #3 | Provider / route Together together | Time to first token0.23 s | Speed measurements
| Endpoint details
|
| #4 | Provider / route Baseten baseten/fp4 | Time to first token0.24 s | Speed measurements
| Endpoint details
|
| #5 | Provider / route SambaNova sambanova | Time to first token0.32 s | Speed measurements
| Endpoint details
|
| #6 | Provider / route Google Vertex google-vertex/global | Time to first token0.38 s | Speed measurements
| Endpoint details
|
| #7 | Provider / route Parasail parasail/fp4 | Time to first token0.39 s | Speed measurements
| Endpoint details
|
| #8 | Provider / route OpenInference open-inference/int8 | Time to first token0.44 s | Speed measurements
| Endpoint details
|
| #9 | Provider / route Weights & Biases wandb/fp4 | Time to first token0.46 s | Speed measurements
| Endpoint details
|
| #10 | Provider / route Nebius nebius/fp4 | Time to first token0.54 s | Speed measurements
| Endpoint details
|
| #11 | Provider / route DeepInfra deepinfra/turbo | Time to first token0.62 s | Speed measurements
| Endpoint details
|
| #12 | Provider / route NovitaAI novita/fp4 | Time to first token0.67 s | Speed measurements
| Endpoint details
|
| #13 | Provider / route Phala phala | Time to first token0.91 s | Speed measurements
| Endpoint details
|
| #14 | Provider / route MARA mara | Time to first token1.07 s | Speed measurements
| Endpoint details
|
| #15 | Provider / route DekaLLM dekallm/bf16 | Time to first token1.11 s | Speed measurements
| Endpoint details
|
| #16 | Provider / route SiliconFlow siliconflow/fp8 | Time to first token1.30 s | Speed measurements
| Endpoint details
|
| #17 | Provider / route Mancer mancer/fp4 | Time to first token1.56 s | Speed measurements
| Endpoint details
|
| #18 | Provider / route Amazon Bedrock amazon-bedrock | Time to first token2.07 s | Speed measurements
| Endpoint details
|
| #19 | Provider / route DigitalOcean digitalocean | Time to first token3.42 s | Speed measurements
| Endpoint details
|
| #20 | Provider / route DeepInfra deepinfra/bf16 | Time to first token8.13 s | Speed measurements
| Endpoint details
|
| — | Provider / route Amazon Bedrock amazon-bedrock/eu-west-1 | Time to first tokenInsufficient recent data | Speed measurements
| Endpoint details
|
Endpoint data fetched .
gpt-oss-120b endpoint guide
20 of 21 gpt-oss-120b endpoints currently have the published data required for this ranking. Groq leads at 0.14 s via groq. The table keeps unranked routes visible so missing measurements do not look like missing provider availability.
The first-token ranking orders endpoints by the recent median time before output begins. This is useful for interactive applications, but it should be read alongside throughput because the first provider to start is not always the first to finish a long answer.
Cerebras currently ranks second at 0.23 s. Compare that gap with input and output price, response-start time, output speed, 30-minute uptime, context length, and the exact route before deciding whether first place is meaningful for your workload.
gpt-oss-120b has 21 published provider options, and performance data is matched to each exact OpenRouter routing tag. Different quantization, context, regional deployment, or provider configuration can change price and behavior even when the underlying model name is identical.
