Skip to main content

​
General

Routeway is a unified API gateway that gives you access to 200+ AI models — from OpenAI, Anthropic, Google, Meta, DeepSeek, MoonshotAI, and more — through a single endpoint and a single API key.Instead of managing separate accounts, SDKs, and billing across every provider, you point your existing OpenAI-compatible client at https://api.routeway.ai/v1 and swap models by changing a single string.
When you call providers directly, you get one model per account, separate billing, separate keys, and separate SDKs to manage. With Routeway:
  • One API key and one base URL for everything
  • Switch between 200+ models without changing your client code
  • Single dashboard for usage, spend, and key management
  • Access models from providers you may not have a direct account with
Yes. Routeway is fully OpenAI-compatible. You can use the official OpenAI Python or Node.js SDK — just change base_url to https://api.routeway.ai/v1 and swap your key. No other code changes required.Anthropic-style requests (/v1/messages) are also supported for models that expose that interface.
Browse the full list at routeway.ai/models or fetch it live from the API:
curl https://api.routeway.ai/v1/models \
  -H "Authorization: Bearer $ROUTEWAY_API_KEY"
Models from OpenAI, Anthropic, Google, Meta (Llama), DeepSeek, MoonshotAI, and others are available. The catalog is updated regularly as new models are released.

​
Account & API Keys

  1. Sign up or log in at routeway.ai/dashboard
  2. Open the API Keys tab
  3. Click Create API Key, give it a name, and confirm
  4. Copy and store the key securely — it won’t be shown again
See Authentication for full setup instructions.
Routeway API keys start with sk- and are 32+ characters long. If your key doesn’t match this format, you may be using a key from a different provider.
Yes. You can create as many keys as you need from the dashboard — for example, separate keys per project, environment, or team member. Each key can be named, monitored, and revoked independently.
Revoke it immediately from the API Keys section of the dashboard and create a new one. Check your recent usage in the Activity tab to see if there were any unexpected requests.
Yes. Key-level controls are available in the dashboard — you can set spending limits and configure fallback behaviour per key. See Budget Controls.

​
Models

Free models have :free appended to their ID (e.g. deepseek-r1:free). They cost $0.00 but are rate-limited to 20 requests/minute and 200 requests/day.Paid (PAYG) models have no suffix and are billed per token with no rate limits at the API level. They require a positive account balance.See Available Models for more detail.
No — free models are intended for development, testing, and experimentation only. Before building anything serious on them, keep in mind:
  • They can be removed at any time. Free model availability depends on provider capacity and our ability to continue offering them. A model that exists today may be gone tomorrow without notice.
  • They can be unstable. Free tiers are lower priority in our routing stack, so you may see higher error rates, slower response times, and more frequent 502 upstream errors compared to PAYG models.
  • Rate limits are strict and shared. 20 RPM and 200 RPD are hard caps with no way to increase them. Under load, you’ll hit these quickly.
  • No SLA. There are no uptime or latency guarantees on free models.
If your project is moving beyond prototyping, switch to the PAYG equivalent (drop the :free suffix). The per-token cost on most models is low enough that real workloads are still very affordable.
The daily limit (200 RPD) resets at 00:00 UTC each day. The per-minute limit (20 RPM) is a rolling window.
Each model in the /v1/models response includes a capabilities object:
"capabilities": {
  "vision": true,
  "function_call": true,
  "reasoning": true
}
You can also filter by capability on the Models Catalog.
The endpoints array in the model object lists all supported API paths, e.g.:
"endpoints": ["/v1/chat/completions", "/v1/responses", "/v1/messages"]
Not all models support all paths. Image models are only available on /v1/images/generations or /v1/images/edits and won’t appear on chat completion routes.
It means the model is temporarily unavailable for routing, usually due to an issue with the upstream provider. Check back shortly or switch to an alternative model. If a model you rely on has been unavailable for an extended period, reach out on Discord.
Yes, for models that list /v1/messages in their endpoints. Point the Anthropic SDK at https://api.routeway.ai/v1/messages and use your Routeway API key.
We add new models continuously as they are released by providers or requested by our community.To get notified immediately when a new model is added, you can join our Discord Server and check the #models channel.

​
Billing & Pricing

Routeway uses Pay-As-You-Go pricing — you’re charged only for the tokens you process, with no monthly commitment. Costs are split between input tokens (prompt + context) and output tokens (generated response), billed per 1 million tokens at rates that vary by model.See How Billing Works for full details and cost examples.
Tokens are the chunks of text AI models process. In English, 1 token is roughly 4 characters or 0.75 words. A typical paragraph is around 100 tokens.As a rough guide:
  • 1,000 tokens ≈ 750 words
  • 1M tokens ≈ a full-length novel
Exact tokenisation varies by model. Most providers publish tokeniser tools you can use to measure a specific prompt.
No. Routeway is strictly Pay-As-You-Go — there are no pre-paid credit bundles to buy. Your account is charged based on actual usage, and you only need a valid payment method on file.
Payments are processed by Polar.sh (fully PCI-compliant). Accepted methods:
  • Credit and debit cards (Visa, Mastercard, Amex, and more)
  • Automatic invoicing
Routeway never stores or touches your card details directly.
PayPal is supported, though it isn’t available by default. For other payment methods, availability may vary — reach out to us and we’ll let you know what’s possible:
Subscriptions give you a reserved monthly or biweekly DeepSeek token budget plus free daily model requests across all plans — useful if you want a predictable DeepSeek allowance.
PlanBiweeklyMonthly
Core$6.99 · 15M tokens$15.99 · 35M tokens
Core Pro$12.99 · 28M tokens$34.99 · 80M tokens
Core Pro also includes a 5% discount on all PAYG text usage. See Billing.
Usage automatically continues on PAYG unless you’ve disabled key fallback in the dashboard. You can turn this off per-key to hard-cap DeepSeek spend.
It depends on the model. Some models offer cache reads at a discounted rate with no write charge (e.g. Kimi K2.7 Code). Others bill both reads and writes, with write rates varying by TTL (e.g. Anthropic models: 5-minute and 1-hour tiers).Some models have no caching at all. Check pricing.caching in the model’s API response, or see the Billing — Prompt Caching section for examples.
Set a hard spending limit from the dashboard. Once your monthly threshold is hit, the API returns 429 errors for that key until the limit resets or is raised. See Budget Controls.
The dashboard shows real-time usage, token breakdowns by model, and historical spend. You can also query it programmatically via the Account API endpoints (/v1/account/balance, /v1/account/activity).
Because Routeway is Pay-As-You-Go and you only pay for usage already consumed, refunds are not available for processed tokens. If you believe you were charged in error, contact support at help@team.routeway.ai or on Discord.

​
API & Integration

EndpointPurpose
POST /v1/chat/completionsText generation (OpenAI-compatible)
POST /v1/responsesOpenAI Responses API
POST /v1/messagesAnthropic-style messages
POST /v1/images/generationsImage generation from a text prompt
POST /v1/images/editsImage editing with a mask and prompt
GET /v1/modelsList all available models with pricing
Not every model is available on every endpoint — check the model’s endpoints array.
A 401 means your API key is missing, invalid, or revoked. Check that:
  • You’re passing the key as Authorization: Bearer <your-key>
  • The key starts with clsk- and was copied in full
  • The key hasn’t been deleted in the dashboard
See Authentication and Error Handling.
A 429 means you’ve hit a rate limit or spending cap:
  • Free models: 20 RPM / 200 RPD limit reached — wait for the reset or switch to a PAYG model
  • Hard spending limit: your monthly cap was hit — raise or remove it in the dashboard
  • Key-level limit: a per-key restriction was triggered — check your key settings
See Rate Limits.
A 502 means the upstream AI provider returned an error or timed out. This is usually temporary. Retry with exponential backoff, or switch to an equivalent model from a different provider. If the issue persists, check Discord for status updates.
Yes. Pass "stream": true in your request body. The response will be delivered as server-sent events (SSE), identical to OpenAI’s streaming format. To receive usage data in a streaming response, add "stream_options": {"include_usage": true}.
Yes, for models where capabilities.function_call is true. The request and response format is identical to OpenAI’s tool-calling spec.
Yes. Any framework that supports a custom OpenAI base URL will work. Set base_url (or equivalent) to https://api.routeway.ai/v1 and use your Routeway API key. No other changes needed.
Yes. See the Integrations page for step-by-step setup guides for Lobe Chat, JanitorAI, ChubAI, Cline, Roo Code, and more.

​
Still stuck?

Can’t find what you’re looking for? Reach out: When reporting an issue, include the trace_id from the error response — it lets us find your request instantly.