Migrate from Together AI to Railwail

TL;DR — Switch in Under 5 Minutes

• Both APIs are OpenAI-compatible — change only the base URL and API key
• All major Together models mirrored: Llama 3.3, Mixtral, DeepSeek V3, Qwen 2.5
• EU-hosted endpoint, EUR billing — clean for European entities
• Plus access to GPT-4o, Claude, Gemini, Flux through the same key
• Comparable per-token pricing on the open-source models

Why Move Off Together AI?

Together AI is great for hosted open-source LLMs at competitive prices. The trade-offs: US-hosted by default, USD-only billing, and the catalog is limited to open-source models. If you want Claude or GPT-4o in the same codebase, you still need separate accounts. Railwail unifies all of these under one API, one EUR invoice, and EU residency.

Step 1 — Get a Railwail API Key

Sign up at railwail.com and create a key in Dashboard → API Keys.

Step 2 — Change the Base URL

TypeScript / JavaScript

Before (Together AI):

import OpenAI from "openai";

const together = new OpenAI({
  apiKey: process.env.TOGETHER_API_KEY,
  baseURL: "https://api.together.xyz/v1",
});

const res = await together.chat.completions.create({
  model: "meta-llama/Llama-3.3-70B-Instruct-Turbo",
  messages: [{ role: "user", content: "Hello" }],
});

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: "llama-3.3-70b-instruct",
  messages: [{ role: "user", content: "Hello" }],
});

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="llama-3.3-70b-instruct",
    messages=[{"role": "user", "content": "Hello"}],
)
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": "llama-3.3-70b-instruct",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

Step 3 — Update Model IDs

Together AI uses fully-qualified Hugging Face style model IDs like meta-llama/Llama-3.3-70B-Instruct-Turbo. Railwail normalises these to short slugs (llama-3.3-70b-instruct). The long IDs are also accepted as aliases for compatibility — your existing Together strings will work, but the short forms are recommended.

API Endpoint Mapping

Model Mapping

Pricing Comparison (per 1M tokens, May 2026)

Why Railwail Over Together AI

• EU billing in EUR with VAT receipts
• Frankfurt-region hosting for GDPR-compliant logs
• Same OpenAI-compatible API — drop-in replacement
• Access to Claude, GPT-4o, Gemini, Mistral La Plateforme through the same key
• Built-in playground at railwail.com/models for A/B testing
• Comparable pricing on Together's lineup, plus per-key spend caps and rate limits

FAQ

Are the Turbo variants (FP8) supported?

Yes. Railwail serves the same FP8-quantised Llama 3.x Turbo variants Together AI offers. Specify -turbo in the model slug to opt into FP8.

What about Together's batch inference?

Use Railwail's POST /v1/batches — same 50% discount on async workloads.

Does the rerank API work the same?

Yes. POST /v1/rerank accepts the same query/documents/top_n payload.

Can I bring my own fine-tuned Llama?

Custom fine-tunes are not yet hosted on Railwail. You can keep them on Together AI and use Railwail for everything else.

How is latency vs Together AI?

First-token latency from EU origins is typically 30-80ms faster via Railwail's Frankfurt edge than via Together's US-default routing.

Next Steps

• Create your Railwail account at railwail.com
• Generate an API key in Dashboard → API Keys
• Change baseURL to https://api.railwail.com/v1
• Update model IDs to Railwail short slugs (or keep long Together IDs as aliases)
• Read the full reference at railwail.com/docs
• Compare per-token pricing at railwail.com/pricing