Start typing to search.
Ranks providers by how quickly the first generated token appears. Providers without enough recent data remain visible after ranked rows.
Weights & Biases currently ranks first at 0.16 s via wandb/bf16.
| Rank | Provider / route | Time to first token | Speed measurements | Endpoint details |
|---|---|---|---|---|
| #1 | Provider / route Weights & Biases wandb/bf16 | Time to first token0.16 s | Speed measurements
| Endpoint details
|
| #2 | Provider / route Groq groq | Time to first token0.29 s | Speed measurements
| Endpoint details
|
| #3 | Provider / route DeepInfra deepinfra/bf16 | Time to first token0.33 s | Speed measurements
| Endpoint details
|
| #4 | Provider / route NovitaAI novita/fp8 | Time to first token0.46 s | Speed measurements
| Endpoint details
|
| #5 | Provider / route DeepInfra deepinfra/fp8 | Time to first token0.48 s | Speed measurements
| Endpoint details
|
| #6 | Provider / route Cloudflare cloudflare/fp8 | Time to first token0.52 s | Speed measurements
| Endpoint details
|
Endpoint data fetched .
Llama 3.1 8B Instruct endpoint guide
6 of 6 Llama 3.1 8B Instruct endpoints currently have the published data required for this ranking. Weights & Biases leads at 0.16 s via wandb/bf16. 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.
Groq currently ranks second at 0.29 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.
Llama 3.1 8B Instruct has 6 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.
