No markup on tokens. Just a flat subscription.

Give every team an AI budget. The gateway enforces it.

TokenRouter is an OpenAI-compatible gateway in front of 13 providers. Set budgets and rate limits per team, product, or API key — requests stop the moment a cap is hit, not after the invoice arrives.

Start 14-Day Trial View the Docs

Incoming Prompt:

GPT-5

$5/1M tokens

Claude Opus 4.5

$5/1M tokens

Gemini 2.5 Pro

$2/1M tokens

Routed to GPT-5 - Most cost-effective for this prompt

Budgets

Daily or monthly limits per team, member, or API key.

Hard Caps

Caps, rate limits, and quotas enforced before the request goes out.

One API, 13 Providers

OpenAI-compatible gateway. Route by cost, latency, or quality.

Full Attribution

Every request logged with its cost, tokens, latency, and model.

A drop-in replacement. Two lines to switch.

Keep the OpenAI SDK you already use. Change the base URL and the key — nothing else in your code moves.

Python

from tokenrouter import TokenRouter

client = TokenRouter(api_key="tr_...")

response = client.responses.create(
    model="auto",
    input="How can tokenrouter.io help me?"
)

print(response.output_text)

TokenRouter SDK

OpenAI SDK

Install TokenRouter

pip install tokenrouter

Change one import

Replace openai with tokenrouter

Make your first request

Name any model you already use, or set model="auto" to let the router pick the cheapest one that can handle the job

Compatible with existing OpenAI code

No breaking changes required

Know where every dollar went.

One console for every request across every provider: what it cost, how many tokens, how long it took, and which team spent it. Built for the person who has to answer for the bill.

Your Keys, Your Rates

Bring your own provider keys. You pay OpenAI and Anthropic their prices — we never add a cent per token.

Smart Routing

Simple requests go to cheap models, hard ones to smart models. Or pin a model and we just route it.

Spend Controls

Budgets, hard caps, and rate limits enforced at the gateway — overspend gets blocked, not reported.

Live Analytics

Spend, latency, and error rates by team, key, and provider — live, not at the end of the month.

Spend Controls

Budgets that actually stop overspend.

Most tools tell you about overspend after the invoice. TokenRouter checks the budget before the request leaves the gateway.

Hard caps that hold

Set daily or monthly budgets per team, member, or API key. When a budget runs out, requests are blocked — not flagged.

Alerts before the cap

Owners get notified at 50, 80, and 100% of budget, so a blocked request is never a surprise.

Rate limits & quotas

Per-key request and token limits keep one runaway script from eating the month's budget in an afternoon.

One bill, fully explained.

Every request is metered and attributed to a team, a member, and a key. When finance asks what the AI line item was, you have the answer down to the request.

Budgets & Hard Caps
Spend Alerts
Team-Level Attribution
Monitor in Real Time

Thirteen providers behind one API.

OpenAI, Anthropic, Google, Meta, Mistral, DeepSeek, and more. Add your provider keys once, then switch models — or mix them across teams — without touching application code.

Watch spend as it happens

Cost, tokens, and latency for every request, the moment it completes — filterable by team, member, and key.

One gateway routing across 13 providers

Drop-in replacement for your favorite SDKs

Your providers, your keys

Connect OpenAI, Anthropic, Google, and ten more. You keep the provider relationship and their pricing — we route, meter, and enforce.

Put a hard limit on your AI spend this week. Flat subscription, zero token markup.

Start 14-Day Trial