Create and store the key
Create the key in Console and load it from AIWAVE_API_KEY. Keep it out of source control and client-side code.
Send the smallest request that still proves the route.
Create a key, set the AIWave base URL, pin a verified model ID, and inspect the usage returned with the response.
Four steps from key to evidence.
Use the same model ID and payload shape in the code and in your acceptance note.
Set the base URL
Point the OpenAI-compatible client to https://aiwave.live/v1.
Pin the tested model ID
Start with deepseek-v4-pro, then use the same test pattern for glm-5.1 or kimi-k3.
Send one representative request
Check the response, status, usage fields, request record, and charge before you move more traffic.
The key is shared. The request contract is not.
Use the protocol that belongs to the selected family. This prevents a model ID from being sent to a path that returns a misleading 404 or an unreadable response envelope.
Route matrix checked 2026-09-10| Family | Route | Protocol | Boundary |
|---|---|---|---|
| GPT | /v1/chat/completions | OpenAI-compatible | Route facts recorded; public price pending |
| Chinese models | /v1/chat/completions | OpenAI-compatible | Public rate card available |
| Claude | /v1/messages | Anthropic native | Native route; compatible path is model-dependent |
| Gemini | /v1beta/models/{id}:generateContent | Gemini native | Native route; image IDs do not use /v1/images/generations |
| Image routes | /v1/images/generations or provider-native | Route-specific | Check the model-family contract before sending media |
See Models for exact IDs and Request Recovery when the route and model do not agree.
Start from the client you already use.
Each example uses the same verified route and model ID.
Python
import os
from openai import OpenAI
client = OpenAI(base_url="https://aiwave.live/v1", api_key=os.environ["AIWAVE_API_KEY"])
response = client.chat.completions.create(model="deepseek-v4-pro", messages=[{"role": "user", "content": "Hello"}])
print(response.usage)Node.js
import OpenAI from "openai";
const client = new OpenAI({ baseURL: "https://aiwave.live/v1", apiKey: process.env.AIWAVE_API_KEY });
const response = await client.chat.completions.create({ model: "deepseek-v4-pro", messages: [{ role: "user", content: "Hello" }] });
console.log(response.usage);curl
curl "https://aiwave.live/v1/chat/completions" \
+ -H "Authorization: Bearer $AIWAVE_API_KEY" \
+ -H "Content-Type: application/json" \
+ -d '{"model":"deepseek-v4-pro","messages":[{"role":"user","content":"Hello"}]}'Before you expand.
Compare the dated rate, response, usage, and completed charge for one representative request before adding tools, retries, or production traffic.
Compare the dated rate, response, usage, and completed charge for one representative request before adding tools, retries, or production traffic.
Bring your agent
Keep the agent. Change the model route.
OpenClaw, WorkBuddy, and a custom agent can use the same AIWave key and USD balance. Start with an OpenAI-compatible model and exact model ID, then use the native route when the model family requires it.