One API for every chatbot you ship

Works with every messaging platform

Any platform that supports HTTP webhooks works — the AIPower API is the LLM layer, not the messaging layer.

The chatbot pattern — streaming replies

Stream tokens to the user as they arrive. Works identically across all 16 models.

// Node.js — works inside your webhook handler (WhatsApp / Telegram / etc.)
import OpenAI from "openai";

const aipower = new OpenAI({
  baseURL: "https://api.aipower.me/v1",
  apiKey: process.env.AIPOWER_API_KEY,
});

async function handleUserMessage(userId: string, text: string) {
  const stream = await aipower.chat.completions.create({
    model: "auto",  // DeepSeek V3 by default — 91% cheaper than GPT-5.4
    stream: true,
    user: userId,   // Tag requests per-user for billing + analytics
    messages: [
      { role: "system", content: "You are a friendly assistant." },
      ...conversationHistory[userId],  // Your session store
      { role: "user", content: text },
    ],
  });

  let full = "";
  for await (const chunk of stream) {
    const delta = chunk.choices[0]?.delta?.content || "";
    full += delta;
    await sendToPlatform(userId, delta);  // WhatsApp / Telegram / etc.
  }
  conversationHistory[userId].push(
    { role: "user", content: text },
    { role: "assistant", content: full },
  );
}

Pick the right model for your chatbot

Protect against abuse & runaway cost

Tag each request with a user ID. Set daily spending caps. Bail when users hit their quota.

// Option 1: Account-wide cap
// Set daily_cap_cents=500 in your dashboard → auto-halt at $5/day

// Option 2: Per-user cap (your app-side)
const userDailySpend = await db.getSpendToday(userId);
if (userDailySpend > 50) {  // 50 cents = ~150k free-tier tokens
  return "You've hit your daily limit. Upgrade to continue.";
}

// Option 3: Per-user analytics
const res = await aipower.chat.completions.create({
  model: "auto",
  user: userId,  // Tag it — query /api/usage/logs?user=...
  messages: [...],
});

Your chatbot never goes down from upstream

Chatbot FAQ

Building something else?