Migrate from Google AI Studio / Vertex AI to Railwail

TL;DR — Switch in Under 10 Minutes

• Drop @google/generative-ai and google-cloud-aiplatform — use the OpenAI SDK instead
• No GCP project, no service account JSON, no IAM roles — just an API key
• Gemini 2.5 Pro, Flash, Flash-Lite all available through Chat Completions
• 1M context window preserved; multimodal (image, audio, video) inputs supported
• EU-hosted endpoint, EUR billing, plus 274 other models behind one API key

Why Move Off Google AI Studio / Vertex AI?

Google's Gemini is one of the most capable multimodal models on the market — but accessing it via Vertex AI requires a GCP project, service account credentials, IAM bindings, and a billing account in USD. AI Studio is friendlier but rate-limited and not production-ready. Railwail exposes the same Gemini models through the standard OpenAI Chat Completions API with no GCP setup whatsoever.

Common migration triggers: avoiding GCP project sprawl, escaping AI Studio's free-tier RPM caps, EUR billing for European subsidiaries, and unifying Gemini calls with GPT-4o or Claude under one SDK.

Step 1 — Get a Railwail API Key

Create an account at railwail.com and generate a key in Dashboard → API Keys. No GCP project required.

Step 2 — Replace the Google SDK

TypeScript / JavaScript

Before (Google AI Studio):

import { GoogleGenerativeAI } from "@google/generative-ai";

const genAI = new GoogleGenerativeAI(process.env.GOOGLE_API_KEY);
const model = genAI.getGenerativeModel({ model: "gemini-2.5-pro" });

const result = await model.generateContent("Hello Gemini");
console.log(result.response.text());

import OpenAI from "openai";

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

const res = await client.chat.completions.create({
  model: "gemini-2.5-pro",
  messages: [{ role: "user", content: "Hello Gemini" }],
});
console.log(res.choices[0].message.content);

Python

from openai import OpenAI

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

resp = client.chat.completions.create(
    model="gemini-2.5-pro",
    messages=[{"role": "user", "content": "Hello Gemini"}],
)
print(resp.choices[0].message.content)

cURL

curl https://api.railwail.com/v1/chat/completions \
  -H "Authorization: Bearer $RAILWAIL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-2.5-pro",
    "messages": [{"role": "user", "content": "Hello Gemini"}]
  }'

Step 3 — Multimodal Inputs (Images, Audio, Video)

Gemini's killer feature is native multimodality. On Railwail, multimodal inputs use the standard OpenAI vision schema (content array with image_url or input_audio blocks). Railwail translates to Gemini's native parts format internally.

const res = await client.chat.completions.create({
  model: "gemini-2.5-pro",
  messages: [{
    role: "user",
    content: [
      { type: "text", text: "What is in this image?" },
      { type: "image_url", image_url: { url: "https://example.com/img.jpg" } }
    ]
  }],
});

Step 4 — Function Calling / Tool Use

Gemini's function calling translates to the OpenAI tools schema. Define your tools as JSON Schema, set tool_choice, and read tool_calls off the response — Railwail handles the conversion to Gemini's tools.functionDeclarations format and back.

API Endpoint Mapping

Gemini Model Mapping

Pricing Comparison (per 1M tokens, May 2026)

Why Railwail Over Google AI Studio / Vertex

• No GCP project — just an API key
• EUR billing with VAT receipts — clean accounting for EU entities
• OpenAI-compatible SDK — unify Gemini calls with GPT-4o, Claude, Llama in one client
• EU-hosted gateway — Frankfurt region for GDPR-friendly logs
• Higher default rate limits than AI Studio's free-tier
• 1M context, multimodal, function calling — all preserved
• Built-in playground to A/B Gemini vs other models on the same prompt

FAQ

Do I lose Gemini's 1M context window?

No. The full 1M-token context is available on gemini-2.5-pro through Railwail. Tiered pricing above 200k tokens is also passed through.

What about Gemini's grounding with Google Search?

Search grounding is a Vertex-specific feature that requires GCP-side configuration. Railwail does not currently expose it directly. For grounded responses, use Perplexity Sonar models on Railwail (perplexity-sonar-large), which provide similar real-time search functionality.

Can I do batch generation?

Yes. Railwail's POST /v1/batches accepts the OpenAI batch format for any Gemini model, with the same 50% discount on async workloads.

Is Gemini Live / real-time streaming supported?

Standard token streaming via SSE is supported. Gemini Live's bidirectional audio-stream API is on the roadmap — track railwail.com/changelog for updates.

What about safety settings?

Gemini's safetySettings field is supported via the metadata.safety_settings extension parameter. Default thresholds match Google AI Studio's defaults.

Where is the data processed?

Railwail routes Gemini calls through Google's EU endpoint when you call from an EU origin, keeping inference data in-region. Request logs are stored in Frankfurt.

Next Steps

• Create your Railwail account at railwail.com
• Generate an API key in Dashboard → API Keys
• Replace @google/generative-ai with the OpenAI SDK in your code
• Update baseURL to https://api.railwail.com/v1
• Read the full API reference at railwail.com/docs
• Browse Gemini models at railwail.com/models?provider=google
• Compare per-token pricing at railwail.com/pricing