Home › API Gateway

API Gateway

One API. Every Major Model. Smarter Routing. Lower Costs.

APICuts routes your LLM requests through a unified API gateway, automatically finding cost-effective routes while preserving your required model, quality, and reliability.

Switch in one line
client = OpenAI(
    base_url="YOUR_APICUTS_ENDPOINT",  # the only line you change
    api_key="YOUR_APICUTS_KEY",
)

Why APICuts?

Lower effective API costs

At least 20% lower all-in cost per request, after our fee, or you pay nothing. See pricing.

One endpoint, minimal code changes

Change one base URL. Your prompts, tools and code stay the same. See the quickstart.

Intelligent routing and optimization

The lowest-cost route to a successful result, plus caching, batching, and prompt and output work. See routing.

Automatic failover

If a provider errors or slows down, eligible requests move to a healthy route. See reliability.

Secure, transparent API access

Authorized providers only, prompts and outputs not stored by default, and an EU or US data region. See security.

Real-time spending and savings analytics

Actual spend, baseline costs and realized savings, tracked per token, live. See the savings dashboard.

OpenAI, Claude, Gemini, DeepSeek and more

Every major provider, cloud platform and authorized inference platform, plus models you host. See where traffic goes.

What's automatic and what we set up

Automatic in the gateway
  • Provider routing and failover
  • Prompt caching and cache-aware routing
  • Semantic caching, on routes where you switch it on
  • Budgets, usage limits and alerts
  • The live savings dashboard
Set up with you by our engineer
  • Your routing rules and data region
  • Prompt and context trimming
  • Output limits
  • Model-routing tests against your evaluations
  • Which jobs move to batch processing
  • Retry and agent-loop limits

Once they're set up, these changes run in the gateway on every request.

Switch in three steps

  1. Get your endpoint. After a free teardown, we set up your apicuts endpoint and key.
  2. Change one line. Point your SDK's base URL at apicuts. It's OpenAI-compatible, so your prompts, tools and code stay the same.
  3. Watch the savings live. Cost, savings and latency show up per token in your dashboard.

Copy-and-paste code for Python, TypeScript, LangChain, LlamaIndex and the Vercel AI SDK is in the quickstart and integration guides.

Routing

Intelligent provider routing

Each request goes to the cheapest suitable provider for the model you asked for.

Optional intelligent model routing

Lower-cost equivalent models are used only where you approve them, after they pass your quality evals.

Automatic failover

If a provider fails, eligible requests move to a healthy route, so production keeps running.

Verified, authorized model access

Requests go only to verified, authorized upstream providers.

Cost

Lower effective API costs

At least 20% lower cost per equivalent request, after our fee, or you pay nothing.

Caching and batch optimization

Provider prompt caching for repeated prefixes, optional semantic caching, and batch processing for work that can wait.

Budgets, usage limits and alerts

Set budgets, usage limits and alerts, so there are no unexpected bills.

Transparent pricing dashboard

Cost, savings and cost per request for every model, tracked per token, live.

Control and security

BYOK (Bring Your Own Key)

Keep your existing provider accounts and keys.

Zero data retention by default

Prompts and outputs pass straight through and are not stored. Only request metadata is logged. Semantic caching is opt-in per route, and its entries expire.

EU or US data region

Choose where your requests are processed.

Request metadata

Model, provider, latency, tokens and cost for every request, never its content.

Same model quality and capabilities

You get the model you asked for, with its full quality and capabilities.

Production

Reliable uptime

The gateway runs on Cloudflare's global network, with automatic failover between providers. A public status page is coming soon.

Enterprise SLAs and support

SLAs and a dedicated optimization engineer for production workloads.

No long-term contracts

Leave anytime: point your base URL back at your provider.

Supported models

OpenAI, Anthropic (Claude), Google Gemini, DeepSeek, Qwen and more. See every supported provider.

How your data flows

Your appSame SDK, one new base URL
apicuts gatewayRouting, caching, batching, failover
Your model providersOpenAI, Anthropic, Google and others

Prompts and outputs pass straight through and are not stored by default. Every vendor that touches your data is listed in our Trust Center.

Want us to find your 20%?

Book a free 15-minute API cost teardown. Bring last month's provider usage and we'll show you your biggest waste and the first fix we'd ship.

Get a Free 15-Minute API Cost Teardown