Will this lock us into another vendor?

No. The router is Apache-2.0 open source, your provider keys stay yours, and the API is OpenAI-compatible, so leaving is a change of base URL.

How much can we save on our LLM bill?

On the public benchmark the router saved 89% against sending everything to the reference model. Import last month's provider bill on the bill page and the estimate is computed on your own traffic before you change a line of code.

How do we compare AI gateways and routers?

The compare pages score each product against the same documented capabilities and say when the review was done; the benchmark page shows measured cost and quality, including the cases the router lost.

90 days of the enterprise, pro and team plans free, no card needed. See the plans

OpenSmartRoute

Providers

One endpoint, every provider behind it.

Every answer from this deployment comes from one of 10 providers - the cloud accounts and self-hosted servers behind 19 of its 85 targets. Prices are per million tokens: the vendor's list price where the model is recognised (6 of 19), otherwise what the target declares. Send model: "auto" and the router picks among them; name a target to bypass routing.

Providersconnected to this deployment: 10
Executable targetsof 85 routed · answer here: 19
Requestslast 7 days · this deployment: 580

Bring the keys you already pay for
Connect an OpenAI, Azure, Anthropic or self-hosted endpoint in the dashboard. Keys are stored encrypted and never shown again.
The router spends each where it is cheapest
Every request is priced across the providers that could answer it; policy runs first, then the best value wins and a failing provider drops down the ranking.
Every answer names who answered and what it cost
Usage, the savings ledger and the trace carry the provider, the upstream model and the price - per request, key, tenant and workspace.

Connected providers

The accounts and servers this deployment answers from right now, with the models each serves and the traffic it took this week.

azure
Azure OpenAI account
unreachable
Models
0
Requests · 7d
0
checked 17 d ago
ollama
Self-hosted server
healthy · 547 ms
Models
0
Requests · 7d
0
checked 17 d ago

One key, every account

Where this deployment's requests went this week.

Executable targets

Each row is a routing target the OpenAI-compatible proxy can answer with, the upstream model it runs as and what a million tokens cost. Targets without a provider still take part in decisions, plans and quotes; see all routing targets.

Target	Kind	Provider	Input / 1M	Output / 1M	Context	Capabilities	Health	Requests · 7d
O Small writing model ollama-qwen3-5-4b/qwen3.5:4b	llm	ollama-qwen3-5-4b	$0.05	$0.05	-	in memory	-	464
B

Programmatic access: GET /api/v1/providers (no key required), GET /v1/models with a key. Operators manage providers on the console; see the platform guide.

Your keys, your providers

Connect the provider accounts you already pay for, and the router spends them where each is cheapest.

A workspace can bring its own keys for any provider on this page; they are stored encrypted, never shown again, and every request that runs on them appears in your usage with the provider that answered and what it cost. Point the OpenAI-compatible endpoint at a target, and the upstream behind it can change without your code noticing.

Photo: Brett Sayles

For the business

What it does for the vendor strategy.

One URL, every vendor
No vendor lock-in
Point your OpenAI-compatible client at one address; add, swap or retire providers in the catalogue without touching application code.
Every vendor side by side
89% less
Lower cost per request
On the 150-prompt benchmark the router spent 89% less than sending every prompt to the reference model, and the judged quality went up, not down.
Read the benchmark
Self-hosting is free
Open source, Apache-2.0

Bring the provider keys you already have.

Connect them in a free workspace and the router spends each where it is cheapest for the request - with the provider that answered and what it cost in every trace.

Create a free account How providers connect

One endpoint, every provider behind it.

Bring the keys you already pay for

The router spends each where it is cheapest

Every answer names who answered and what it cost

Connected providers

Where this deployment's requests went this week.

Executable targets

Connect the provider accounts you already pay for, and the router spends them where each is cheapest.

What it does for the vendor strategy.

No vendor lock-in

Lower cost per request

Open source, Apache-2.0

Bring the provider keys you already have.

The request moves to the next-ranked provider, not into an error.