Home › Docs › Choosing an LLM gateway

Choosing an LLM gateway

There are great LLM gateways for many jobs. Here is how the main kinds differ, and when apicuts is the right one.

Looking for an OpenRouter alternative?

Marketplaces like OpenRouter are known for a huge model catalog and pay-as-you-go access to many models through one key. Choose apicuts when your priority is a lower bill: we optimize your traffic for you and guarantee at least 20% lower all-in cost, after our fee.

Full comparisons: APIcuts vs OpenRouter and APIcuts vs Requesty.

Looking for a CometAPI alternative?

Resellers like CometAPI sell model access below official prices. Choose apicuts when you want a guaranteed saving on your total bill, and traffic that goes only to official and authorized providers.

Full comparison: APIcuts vs CometAPI.

Looking for a LiteLLM alternative?

Open-source proxies like LiteLLM give you full control, and you typically run and maintain them yourself. Choose apicuts when you'd rather not operate a gateway, or keep LiteLLM and add apicuts behind it as an OpenAI-compatible upstream.

Full comparison: APIcuts vs LiteLLM.

Looking for a Portkey alternative?

Production gateways like Portkey and Kong are known for governance, guardrails and observability. Choose apicuts when cost is the problem to solve, with quality protected by your own evaluations.

Full comparison: APIcuts vs Portkey.

Cloud and edge gateways

Cloudflare AI Gateway and Vercel AI Gateway bring caching, analytics and developer integrations close to your stack. apicuts runs on Cloudflare's network too, and adds hands-on optimization and a savings guarantee.

Full comparisons: APIcuts vs Cloudflare AI Gateway and APIcuts vs Vercel AI Gateway.

Observability tools

Tools like Helicone and Langfuse are known for logging, tracing and cost analytics. apicuts shows it too, in your savings dashboard, and then cuts it.

Full comparison: APIcuts vs Helicone.

Smart model routers

Routers like Not Diamond and Martian choose a model per request. apicuts includes model routing, plus caching, batching, prompt work and output limits, measured as the lowest cost per successful task.

Full comparison: APIcuts vs Not Diamond.

Looking for an LLM cost-audit service?

Some services audit your LLM spend and charge a share of the savings they deliver. apicuts charges 10% of verified savings, guarantees at least 20% lower all-in cost after that fee, and goes live in one day.

Five questions to ask any gateway

1. Is the saving guaranteed, and is it measured after the gateway's own fee?
2. How is the saving measured, and is it reconciled with my provider invoices?
3. What protects quality and speed when routes or models change?
4. Where is my data processed, and are prompts and outputs stored?
5. How do I leave?

apicuts' answers: yes, 20% net of our fee; per equivalent request, reconciled with invoices; your evaluations, canary rollouts and rollback; the EU or US region you choose, with no stored prompts or outputs by default; one line of code.