I built an API gateway for 48 AI models — here's how it works

#ai #api #saas #programming

I've been working on RiftAI — an OpenAI-compatible API gateway that gives developers access to 48+ AI models through a single endpoint.

The problem

If you work with multiple AI providers (OpenAI, Anthropic, Google, etc.), you end up managing multiple API keys, different SDKs, and separate billing. It gets messy fast.

The solution

RiftAI provides one API key that works with all major models: Claude, GPT-5, Gemini, DeepSeek, Qwen, GLM, Llama and more. The endpoint is fully OpenAI-compatible, so any existing code that works with OpenAI works with RiftAI — just change the base URL.

Credit-based pricing

Instead of per-token billing, RiftAI uses a daily credit pool. Each model has a cost multiplier based on how resource-heavy it is:

DeepSeek V3: ×1 (very cheap)
Gemini 2.5 Flash: ×2
GPT-5: ×8
Claude Sonnet 4.6: ×25
Claude Opus 4.7: ×60 (premium)

Formula: credits = (input_tokens × multiplier) + (output_tokens × multiplier × 3)

This means you can use cheap models for simple tasks and save credits for premium models when you actually need them.

Quick start

curl https://riftai.su/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-2.5-flash",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

Plans

Free: 100K credits/day, no card required
Paid: from $4.99 to $149.99/mo with up to 100M credits/day

There's a credit calculator on the pricing page so you can estimate costs: riftai.su/pricing

Would love to hear feedback from the dev community.

DEV Community