New Gemini 3 Pro & Claude 4.5 Available

The Backend for Vibe Coding

Power Cursor, Windsurf, and Claude Code with a single key. Access SOTA models without managing multiple env variables.

Cursor

Claude Code

Codex

OpenAI SDK

// ~/.cursor/config.json
{
  "models": [
    "name": "|",
    "url": "https://api.scalellm.dev/v1"
  ],
  "apiKey": "sk_..."
}

7+SOTA Models

99.9%Uptime SLA

<100msGlobal Latency

1 LineTo Migrate

Everything Not To Worry About

Focus on your product. We handle the infrastructure.

Universal Tool Adapter

Works natively with any tool that supports OpenAI or Claude protocols. Switch from Cursor to Windsurf instantly.

Smart Fallbacks

Automatic retries and cross-provider failover. If OpenAI is down, we seamlessly route to Azure or Anyscale.

Zero-Config Context

Drop your API key into .cursor/config and start coding. No complex proxy setup needed. We handle the rest.

Simple, Transparent Pricing

Start small and scale as you grow. No hidden fees.

Free

Try ScaleLLM risk-free.

10 Credits

7-day trial

Access all 7 SOTA models
No credit card required

Start Free Trial

Dev

$5/mo

For hobbyists and side projects.

50 Credits

Access all 7 SOTA models
7-day Analytics History

Get Started

Common Questions

How does pricing work?

You subscribe to a monthly plan and receive API credits. Use these credits to make requests to any supported model. Higher plans offer more credits.

What are API credits?

API credits are our internal billing unit. Each API request consumes credits based on the model and token usage. Check our pricing page for detailed credit costs per model.

Do you charge less for cached tokens?

No, we use simple flat-rate pricing per model. We don't differentiate between cached and non-cached tokens - you pay the same predictable rate for every request.

What happens when I run out of credits?

When your credits are exhausted, API requests will return an error. You can upgrade your plan anytime for more credits. Unused credits reset at the start of each billing cycle.

Which models are supported?

We support the latest SOTA models including Claude Opus 4.5, Claude Sonnet 4.5, Claude Haiku 4.5, Gemini 3 Pro, and Gemini 3 Flash.

Is my data safe?

Absolutely. We act as a passthrough and do not store request bodies or train on your data. All requests are encrypted in transit.

Start building in 5 minutes

Join thousands of developers shipping faster with ScaleLLM.

Start Free Trial