Auto Cache
Enabled by default on supported models. Disable it per request with auto_cache: false or the x-itzi-auto-cache header.
Itzi exposes the formats used by OpenAI and Anthropic clients, plus automatic caching and built-in tools behind one API key.
Authenticate with a Bearer token or x-api-key. Keys are displayed once and stored only as hashes.
/v1/chat/completionsOpenAI Chat Completions/v1/responsesOpenAI Responses/v1/messagesAnthropic Messages/v1/modelsCustomer model catalogStreaming, custom tools, vision and standard tool choice fields follow the client format you already use.
import OpenAI from "openai";
const client = new OpenAI({
apiKey: process.env.ITZI_API_KEY,
baseURL: "https://itzi.app/v1",
});
const stream = await client.chat.completions.create({
model: "gpt-5.6-terra",
messages: [{ role: "user", content: "Summarize this dataset" }],
stream: true,
// Auto Cache is enabled automatically on supported models.
});Use the native API fields where possible, with small Itzi extensions for gateway features.
Enabled by default on supported models. Disable it per request with auto_cache: false or the x-itzi-auto-cache header.
Send standard function tools. Web Search and Python Studio can use native tool types or the itzi_tools extension.
All three generation endpoints support server-sent event streams.
Client errors never expose internal providers. Your Itzi request log still shows tokens, cache state, tools and the amount charged.
Create an account, generate a key and use https://itzi.app/v1 as the base URL.