ANT for Bots & Agents

How it Works

┌──────────┐   ┌─────────────┐   ┌──────────────┐   ┌──────────┐   ┌──────────┐
│ Your Bot │──▶│ Classify LLM│──▶│ Optimize LLM │──▶│ Route AI │──▶│   LLM    │
└──────────┘   └─────────────┘   └──────────────┘   └──────────┘   └──────────┘
                Task type &        Compress           Score &        2,500+ models
                complexity         (optional)         select         30+ providers

Quick Start (2 minutes)

curl -s -X POST https://antbase.ai/auth/register \
  -H 'Content-Type: application/json' \
  -d '{"email": "[email protected]", "password": "your-password"}'

curl -s https://antbase.ai/healthz

Framework Setup

curl -fsSL https://antbase.ai/install.sh | bash

{
  "providers": {
    "antbase": {
      "baseURL": "https://antbase.ai/v1",
      "apiKey": "YOUR_ANT_API_KEY"
    }
  },
  "defaultProvider": "antbase"
}

Virtual Models

Request a virtual model and ANT scores all real models against those preference weights and selects the highest-scoring one.

You can also create custom nests that restrict routing to a specific set of models — see Virtual Models docs.

Bypass routing: Pass any real model name (e.g., anthropic/claude-sonnet-4-6) to skip routing and use that model directly.

Model Routing

ANT's classification LLM analyzes your prompt and routes to the most cost-effective model that can handle it.

Response Headers

ANT adds these headers to every response so you can see what happened:

API Reference

See the full API Reference for detailed request/response schemas.

Pricing

Errors

Multi-Agent Frameworks

Agent Client Protocol (ACP)

FAQ