Gateway & integration
LLM routing and fallback
Route each LLM request by cost, speed or reliability, with bounded fallback to another supplier route for the same model. Set a mode per account, key or request.
tokely.me/llm-routing
LLM gateway: one API for every model
A hosted LLM gateway with one OpenAI-compatible endpoint, account routing modes, bounded fallback, spending limits and request logs. Text, image, video and audio from one balance.
By TokelyReviewed Tokely is a hosted LLM gateway: point an OpenAI-compatible client at https://api.tokely.me/v1, authenticate with one Tokely key and call supported models from several developers. Routing mode, fallback, budgets and request logs are handled by the gateway; you do not deploy or operate a proxy.
Connect in one request
curl https://api.tokely.me/v1/chat/completions \
-H "Authorization: Bearer $TOKELY_API_KEY" \
-H "Content-Type: application/json" \
-H "X-Tokely-Routing: lowest-cost" \
-d '{"model": "kimi-k3",
"messages": [{"role": "user", "content": "Hello"}],
"max_tokens": 256}'Base URL https://api.tokely.me/v1. Create a key in the dashboard and keep it in an environment variable. Quickstart · Model IDs and endpoints
One endpoint, one key, one balance
Supported text, image, video and audio models share one API key and one prepaid USD balance. Switching models is a change of the model ID, not a new vendor account, SDK or invoice. Each model page lists its endpoint, parameters and published rate.
Routing modes you choose per account or request
Balanced, Lowest cost, Fastest, Most reliable and Fixed. Set a default in account settings or send the X-Tokely-Routing header on a request. Lowest cost compares current route quotes for a normalized token mix; Fixed keeps the selected route and fails instead of substituting a model.
Bounded fallback, not blind retries
When a compatible route fails with an eligible error before work is accepted, the gateway can retry another configured route for the same model. Accepted media tasks stay pinned and partial streams are not replayed, so a fallback does not silently duplicate paid work. Some models have only one route.
Spending control built into the gateway
Daily and monthly account budgets are reserved atomically across all keys and the Playground. Keys can have their own limits and model restrictions. A request that would exceed a limit is rejected with a distinct error instead of being billed.
Logs, usage and receipts
Every response carries an X-Request-ID that links to its dashboard log. Logs show attempt outcomes, duration, routing mode and HTTP status. Usage can be exported, signed webhooks report events, and recorded savings use a request-time reference price where a comparable one exists.
What the gateway does not do today
Bring-your-own provider keys (BYOK), team RBAC and SSO, self-hosted deployment, Batch API and speech transcription are not currently offered. Claude models are also available through the native Anthropic Messages API for Claude Code. If your project requires one of them, a self-hosted proxy or another gateway may fit better.
Questions before you switch
What is an LLM gateway?
An LLM gateway is a layer between your application and model providers. It gives you one API and key, chooses a route for each request, handles failures and records usage and cost. Tokely runs it as a hosted service, so there is nothing to deploy.
How is a hosted gateway different from LiteLLM?
LiteLLM is proxy software you deploy and connect to your own provider accounts. Tokely is a hosted API with its own balance and routes: no proxy infrastructure, but also no BYOK. See the LiteLLM comparison for the trade-offs.
Can I control which route a request uses?
Yes, within the published modes. Choose Balanced, Lowest cost, Fastest, Most reliable or Fixed in account settings or with the X-Tokely-Routing header. Supplier credentials and internal channel names are not exposed.
Does the gateway add a platform fee?
No separate platform fee is added on top of the published model rate. You pay the model's Tokely rate from your prepaid balance; payment-network costs are shown at checkout.
START WITH YOUR OWN WORKLOADKnow the price.
Know the price.
Then make the call.
Check the model, compare the rate, and connect with a Tokely key.