AI Fusion
OpenAI-compatible

One endpoint. Many providers.

AI Fusion is a gateway that unifies free-tier and BYOK access across 5 providers behind a single OpenAI-compatible API. Drop in, route intelligently, stay under quota.

5providers
41rotating keys
52models

Built for production routing

Three things every serious gateway needs. Nothing else in the way.

Provider rotation

Atomic key allocation in Redis. Automatic fallback across providers and cooldown on rate-limit responses. You see one endpoint; the gateway does the rest.

Quota + cost control

Per-tenant RPM, monthly request and token caps. Every call is priced against retail rates so you know what you'd have paid elsewhere.

OpenAI-compatible

Point any OpenAI SDK at /v1. Anthropic and Gemini-native surfaces ship too. Live progress streaming via SSE when you want it.

Three lines to switch

If it talks to OpenAI, it talks to AI Fusion.

from openai import OpenAI

client = OpenAI(
    base_url="https://ai.viktorarsov.com/v1",
    api_key="afp_live_…",
)

resp = client.chat.completions.create(
    model="gateway:fast-free",
    messages=[{"role": "user", "content": "Hello, fusion."}],
)

Pricing

Records-only billing. Free tier never expires.

Free

Perfect for personal projects and getting started.

$0/ mo

  • 10000 requests / month
  • 10000000 tokens / month
  • 50 requests / minute
  • 5 providers
  • Free-tier models

Pro

For production workloads requiring scale.

$50/ mo

  • 300000 requests / month
  • 300000000 tokens / month
  • 1500 requests / minute
  • Everything in Starter
  • Premium reasoning models (DeepSeek v3.2, Kimi K2, Qwen3 thinking)
  • Hedged routing
  • Custom guardrails
  • Webhooks
  • SLA support