Start typing to search.
Ranks providers by their recent typical output speed in tokens per second. Providers without enough recent data remain visible after ranked rows.
Cerebras currently ranks first at 857.0 tok/s via cerebras/fp16.
| Rank | Provider / route | Output speed | Speed measurements | Endpoint details |
|---|---|---|---|---|
| #1 | Provider / route Cerebras cerebras/fp16 | Output speed857.0 tok/s | Speed measurements
| Endpoint details
|
| #2 | Provider / route Groq groq | Output speed392.5 tok/s | Speed measurements
| Endpoint details
|
| #3 | Provider / route Baseten baseten/fp4 | Output speed230.0 tok/s | Speed measurements
| Endpoint details
|
| #4 | Provider / route MARA mara | Output speed166.0 tok/s | Speed measurements
| Endpoint details
|
| #5 | Provider / route Amazon Bedrock amazon-bedrock | Output speed154.0 tok/s | Speed measurements
| Endpoint details
|
| #6 | Provider / route Google Vertex google-vertex/global | Output speed146.0 tok/s | Speed measurements
| Endpoint details
|
| #7 | Provider / route SambaNova sambanova | Output speed130.0 tok/s | Speed measurements
| Endpoint details
|
| #8 | Provider / route Together together | Output speed104.0 tok/s | Speed measurements
| Endpoint details
|
| #9 | Provider / route DeepInfra deepinfra/turbo | Output speed103.0 tok/s | Speed measurements
| Endpoint details
|
| #10 | Provider / route Nebius nebius/fp4 | Output speed93.0 tok/s | Speed measurements
| Endpoint details
|
| #11 | Provider / route Parasail parasail/fp4 | Output speed85.0 tok/s | Speed measurements
| Endpoint details
|
| #12 | Provider / route Phala phala | Output speed71.0 tok/s | Speed measurements
| Endpoint details
|
| #13 | Provider / route NovitaAI novita/fp4 | Output speed45.0 tok/s | Speed measurements
| Endpoint details
|
| #14 | Provider / route DeepInfra deepinfra/bf16 | Output speed39.0 tok/s | Speed measurements
| Endpoint details
|
| #15 | Provider / route OpenInference open-inference/int8 | Output speed38.0 tok/s | Speed measurements
| Endpoint details
|
| #16 | Provider / route Weights & Biases wandb/fp4 | Output speed38.0 tok/s | Speed measurements
| Endpoint details
|
| #17 | Provider / route Mancer mancer/fp4 | Output speed37.0 tok/s | Speed measurements
| Endpoint details
|
| #18 | Provider / route DekaLLM dekallm/bf16 | Output speed32.0 tok/s | Speed measurements
| Endpoint details
|
| #19 | Provider / route SiliconFlow siliconflow/fp8 | Output speed24.0 tok/s | Speed measurements
| Endpoint details
|
| #20 | Provider / route DigitalOcean digitalocean | Output speed24.0 tok/s | Speed measurements
| Endpoint details
|
| — | Provider / route Amazon Bedrock amazon-bedrock/eu-west-1 | Output speedInsufficient 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. Cerebras leads at 857.0 tok/s via cerebras/fp16. The table keeps unranked routes visible so missing measurements do not look like missing provider availability.
The output-speed ranking compares recent median generated tokens per second. It favors routes that complete long generations quickly, while first-token latency remains a separate measure of how responsive the endpoint feels at the beginning.
Groq currently ranks second at 392.5 tok/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.
