Quickstart

Visit platform.flex.ai and create an account. Sign-up is self-serve — no approval step and no waiting list. The one exception is a company that has claimed its own email domain and made it invitation-only. If that is yours, sign-up is refused and you need an invite from someone who already administers your organization.

2. Create a workspace

Signing up gives you an organization; a workspace is what holds API keys, usage and logs, and you create the first one yourself from the dashboard. You’ll be asked which region it lives in while you do. That choice is permanent, so if you need your inference to stay in the EU, read Regions before you pick.

3. Create an API key

No card is needed for this step. On Workspace home in the dashboard, find Create a new key, name it, click Create Key, and copy the sk-… value — it’s shown once. What the key does not carry yet is money. Inference draws against your account balance, and a new account’s balance is zero, so your first request comes back 402 Payment Required until you fund it. That is deliberate: you can wire the key into your code, your CI and your secret store before any money is involved, and add billing when you’re ready to send real traffic.

4. Add funds

From the dashboard’s Billing page, add a billing address and a card. We can currently bill addresses in the European Union, the United States and India — if yours is elsewhere, the card cannot be added yet, and support@flex.ai can tell you where that stands. Then top up by any amount between $5 and $500 (€5 to €500 on an EU account). The amount you enter is the credit you get; sales tax or VAT is added on top of it at checkout. Your balance is credited once the payment settles — normally a few seconds, and the dashboard shows the top-up as pending until it does. From then on requests simply start succeeding, with no key regeneration and no restart. Adding a card also moves the key off the free rate-limit tier. See Billing & Quotas for the per-minute limits, how spend is metered, and what the 402 body looks like.

You’re billed only for what you use, at per-model rates. Credit — from a program, a negotiated offer, or the verified first top-up match on a US-dollar account — lands in the same balance as the money you top up, so it makes that balance last longer rather than being spent ahead of it. Your card is charged when funds are added, not per request; the match in particular is applied after your first top-up has been charged. See the credits FAQ.

5. Make a request

Pick a text model from the catalog. The examples below use Llama-3.3-70B-Instruct-FP8, and call the US endpoint, https://api.flex.ai/v1. If you chose the EU region in step 2, change the host to https://eu.api.flex.ai/v1. Nothing else moves: same paths, same request bodies, same key header. The two regions are separate stacks rather than two names for one, so a key only works against its own — an EU key sent to api.flex.ai comes back 401, which reads like a bad key rather than a wrong host. The two catalogs are not identical, and which models a region serves is not something you can read ahead of time, so list /v1/models on your own host and take the model id from that answer rather than from the example below. A model id your host does not serve comes back 404, with a close-match hint. Regions has the full picture.

curl https://api.flex.ai/v1/chat/completions \
  -H "Authorization: Bearer $FLEXAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "Llama-3.3-70B-Instruct-FP8",
    "messages": [
      {"role": "user", "content": "In one sentence, what is FlexAI?"}
    ]
  }'

# pip install openai
import os
from openai import OpenAI

client = OpenAI(
    base_url="https://api.flex.ai/v1",
    api_key=os.environ["FLEXAI_API_KEY"],
)

response = client.chat.completions.create(
    model="Llama-3.3-70B-Instruct-FP8",
    messages=[{"role": "user", "content": "In one sentence, what is FlexAI?"}],
)
print(response.choices[0].message.content)

// npm install openai
import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.flex.ai/v1",
  apiKey: process.env.FLEXAI_API_KEY,
});

const response = await client.chat.completions.create({
  model: "Llama-3.3-70B-Instruct-FP8",
  messages: [{ role: "user", content: "In one sentence, what is FlexAI?" }],
});
console.log(response.choices[0].message.content);

The same call against the EU endpoint:

cURL (EU workspace)

curl https://eu.api.flex.ai/v1/chat/completions \
  -H "Authorization: Bearer $FLEXAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "Llama-3.3-70B-Instruct-FP8",
    "messages": [
      {"role": "user", "content": "In one sentence, what is FlexAI?"}
    ]
  }'

You should see a single-line description come back. That’s it — your key works.

Next steps

Stream responses

Deliver tokens as they’re generated instead of waiting for the full response.

Call tools

Let the model invoke your functions and work with their output.

Discover models

Filter the live catalog from code to find the right text, vision, or embedding model.

Understand billing

Rate limits, credit exhaustion behavior, and the 402 response.

Handle errors

Every 4xx/5xx we return, with example bodies and how to recover.

Get Started

Guides

Agents

Reference

Changelog

Chat

Completions

Embeddings

Models

Usage

2. Create a workspace

3. Create an API key

4. Add funds

5. Make a request

Next steps

Stream responses

Call tools

Discover models

Understand billing

Handle errors

​1. Sign up

​2. Create a workspace

​3. Create an API key

​4. Add funds

​5. Make a request

​Next steps

Stream responses

Call tools

Discover models

Understand billing

Handle errors

1. Sign up

2. Create a workspace

3. Create an API key

4. Add funds

5. Make a request

Next steps