Build with Railwail

One API key and one balance for chat, image, video, speech, transcription and embedding models; the rest of the catalog (music, upscaling, background removal …) runs on the model pages. The API is OpenAI-compatible, so the official OpenAI SDKs work with a new base URL. There is also a small npm SDK and an MCP server that gives coding agents the catalog and these docs.

1. 1Create an API keySign in, open API keys, create one. It starts with rw_live_ and is shown once.
2. 2Try a model in the browserEvery model page has a playground and ready-made API code for that model.
3. 3Connect your coding agentClaude Code, Cursor or VS Code read the catalog and docs over MCP.

What you can call#

Every endpoint takes the same key. The last column counts the models each endpoint can run right now; it is computed from the live catalog, so it matches what the API accepts.

Your first request#

Put the key in RAILWAIL_API_KEY (see Quick Start) and run one of these. max_tokens keeps the reserved credits small, which also keeps the call inside the free trial limits.

curl https://railwail.com/api/v1/chat/completions \
  -H "Authorization: Bearer $RAILWAIL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4o-mini",
    "messages": [{"role": "user", "content": "Explain vector databases in two sentences."}],
    "max_tokens": 300
  }'

# pip install openai
import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["RAILWAIL_API_KEY"],
    base_url="https://railwail.com/api/v1",
)

completion = client.chat.completions.create(
    model="gpt-4o-mini",
    messages=[{"role": "user", "content": "Explain vector databases in two sentences."}],
    max_tokens=300,
)
print(completion.choices[0].message.content)

// npm i openai   (ESM: save as .mjs)
import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.RAILWAIL_API_KEY,
  baseURL: "https://railwail.com/api/v1",
});

const completion = await client.chat.completions.create({
  model: "gpt-4o-mini",
  messages: [{ role: "user", content: "Explain vector databases in two sentences." }],
  max_tokens: 300,
});
console.log(completion.choices[0].message.content);

// npm i railwail   (ESM: save as .mjs)
import railwail from "railwail";

const rw = railwail(process.env.RAILWAIL_API_KEY);

const reply = await rw.run("gpt-4o-mini", "Explain vector databases in two sentences.", {
  max_tokens: 300,
});
console.log(reply);

What runs where#

A few of the models each endpoint runs today, with real prices as charged. Every model page has its own playground and API code.

SDKs and tools#

Prices, limits and the free trial#

• Prices are in US dollars; balances are credits (1 credit = USD 0.01). Every model page and the pricing page show the price per token, image, second or run.
• Before a run the API estimates the price and holds it; token and GPU-time runs are settled to real usage afterwards. Failed runs are refunded automatically.
• Each key has its own rate limit (default 600 requests per minute, 60 to 6,000 when you create it). See Rate limits.
• Signing in with Google gives 10 free credits ($0.10), usable 24 hours after sign-up, for runs of up to 2 credits each. Keep max_tokens low while you test on them.