One account for global AI models.
Browse models, create API keys, top up credits and route requests through a single OpenAI-compatible endpoint. Ani Router keeps pricing, usage and team controls in one clean console.
Live gateway
anirouter/v1
Auto route, track spend and switch models without changing app code.
pk_live_82k...
5 models
$38.42
820 ms
Supported by leading AI providers
Model catalog
Find the right model before you call it.
Search by provider, modality, context length and price. Developers can compare models quickly and copy the exact model id into existing SDKs.
Open model catalogEverything a developer needs after the first request.
The public experience is simple for individuals, while the same platform still supports teams, resellers and managed cloud accounts when customers grow.
Model discovery
Browse provider, context, price and capability before integration.
API keys
Create keys for apps, environments and teammates with clear limits.
Credit wallet
Recharge, view balance and connect billing workflows as usage grows.
Usage analytics
See spend, tokens, latency and model share from one dashboard.
Team controls
Approve users, set budgets and restrict model access by team.
Multi-provider routing
Route Claude through Bedrock Runtime, GPT-5 through Bedrock Mantle, and MiniMax-M3 through its official API.
Compatible with familiar SDKs.
Point your OpenAI-compatible client at Ani Router, choose a model id, and keep the rest of your app unchanged.
import OpenAI from "openai";
const client = new OpenAI({
apiKey: process.env.ANIROUTER_API_KEY,
baseURL: "https://demo.octopusrotuer.com/api/proxy",
});
const result = await client.chat.completions.create({
model: "claude-sonnet-4-6",
messages: [{ role: "user", content: "Ship faster." }],
});Simple credit-based pricing.
Users can start with credits, then add team budgets, invoices and private deployment options when needed.
Free start
Create keys and test compatible models.
Pay as you go
Recharge credits and route production traffic.
Team budget
Add approvals, member limits and private model access.