Overview
Three ways to use Elektric.
Benchmark your AI requests, call an exact model with Wire, or let Grid choose the right model for every request.
#Three products
Benchmark
Know before you route.Benchmark analyzes representative AI traffic to compare expected quality and cost across supported models.
Explore Benchmark →Wire
You choose the model.Wire routes your request to the exact public model ID you provide.
Explore Wire →Grid
Elektric chooses the model.Grid selects an eligible model for each request based on quality, cost, and availability.
Explore Grid →#One integration underneath it all.
Wire and Grid share the same API key, SDKs, REST API, usage logs, and billing. Move between explicit model selection and Grid without rebuilding your integration.
#Already using OpenAI?
Keep your existing client for supported chat completions. Change the API key and base URL, then use elektric-grid so Elektric selects the model.
import OpenAI from "openai";
const client = new OpenAI({
apiKey: process.env.ELEKTRIC_API_KEY,
baseURL: "https://elektric.ai/v1",
});
const response = await client.chat.completions.create({
model: "elektric-grid",
messages: [{ role: "user", content: "Hello" }],
});Compatibility covers a focused API surface, not every OpenAI endpoint or option. Read the OpenAI migration guide →
#Build on AI, not on a model.
Models and providers change. Elektric keeps that complexity behind one interface, so your application can adopt new models without being rebuilt around them.