OpenAI-compatible API

Every model.
One key.
Cost + 5%.

Point any OpenAI client at Router and call GPT and Claude models with one key. You pay the provider's own list price plus exactly 5%, settled from the provider receipt. Free open models cost nothing.

Get an API key See prices

No card for free models. Prepay from $20 for paid models. No subscription.

Two lines to switch

# Any OpenAI SDK or tool
export OPENAI_BASE_URL="https://router.writingmate.ai/v1"
export OPENAI_API_KEY="your-router-key"

# Python
from openai import OpenAI
client = OpenAI()
reply = client.chat.completions.create(
    model="anthropic/claude-sonnet-4-6",
    messages=[{"role": "user", "content": "Hello"}],
)

Direct to OpenAIAnthropicGroq No resellers, no fallbacks, no prompt logging.

Pricing

The provider's price, plus 5%. That is the whole price list.

USD per 1M tokens. The Provider column is the official list price. The Router column is what you pay. Prices refresh from the provider price list every few minutes.

Model	Input		Output
Model	Provider	Router	Provider	Router
Loading current prices…

Cached input and cache writes are also billed at provider price plus 5%. Machine-readable list: /v1/models.

Free models

Open models served through Groq. 5 requests per minute, 50 per UTC day, shared capacity. Works with a $0 balance.

Paid models

+5%

Each request reserves its worst case, then settles to the exact usage on the provider receipt. Never an estimate.

Funding

from $20

Prepay by card. Per-request, daily, and monthly caps protect your balance. Unused funds stay in your account.

Quickstart

Up and running in a minute.

1Sign up
Google, Writingmate, or email.
2Create a key
One active key, revoke any time.
3Send a free request
Use free/general or free/coding.
4Add funds
Unlock GPT and Claude models.

curl https://router.writingmate.ai/v1/chat/completions \
  -H "Authorization: Bearer $ROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"free/general","messages":[{"role":"user","content":"Say hello in five words."}]}'

Endpoint	Access	Streaming
/v1/chat/completions	free and paid	Yes
/v1/responses	paid	Yes
/v1/completions	paid	Yes
/v1/models	public	No

Machine-readable matrix: /v1/compatibility.

FAQ

Straight answers.

Why not call the providers directly?: You can. Router gives you one key, one balance, and one bill for every provider, for 5%. No separate accounts, cards, or spend limits per vendor.
How is the 5% calculated?: On the provider's official list price for the tokens on the provider receipt. The price list is checked automatically and the exact rate is shown on every model.
Do you log my prompts?: No. Router forwards requests to the provider and stores token counts and cost for billing only.
What if my balance runs out mid-request?: Each request reserves its worst case before it starts, so a request either runs fully or is refused up front. Free models keep working at $0.
Can I use it with Writingmate?: Yes. Writingmate itself runs on Router, and you can sign in with your Writingmate account.

Start with a free request.

Get an API key Sign in

Every model.One key.Cost + 5%.