Relay: Groq LPUOpenAI-compatible SSEDevnet

Route a prompt through the substation.

Every response streams token by token through the routing gateway. Inference is relayed through Groq LPU capacity until the first bare-metal node registers. The wire protocol does not change.

response streamidle

Awaiting prompt. Every response streams token by token through the routing gateway.

0 tokens

Cmd/Ctrl + Enter to send · Esc to stop

curl https://api.wattz.fi/v1/chat/completions \
  -H "content-type: application/json" \
  -d '{
  "model": "llama-3.1-8b-instant",
  "messages": [
    {
      "role": "system",
      "content": "You are Wattz, a Solana-native inference marketplace assistant."
    },
    {
      "role": "user",
      "content": "You are on a Solana AI inference marketplace called Wattz. Introduce yourself in three concise sentences to a builder who has never heard of it."
    }
  ],
  "temperature": 0.7,
  "max_tokens": 512,
  "stream": true
}'

Metering

Throughput

--tok / s

Model

First token

Stream time

Tokens out

Tokens in

Served via

Attestation

relay

Relay path — TEE attestation activates when the first bare-metal node registers.