First request in two minutes.

1. Create an account

Sign up for a Failure ID at /signup with the invite key the lab sent you (no key? ask for one). Turn on two-factor straight away — see authentication.

2. Create an API key

Then open the Console → Keys and create a key. It is shown once. We store only a hash, so if you lose it you make a new one. Set a monthly spend cap on it while you are there.

export FAILURE_API_KEY="paste-your-key-here"

3. Send a request with curl

curl https://api.failure.fail/v1/chat/completions \
  -H "Authorization: Bearer $FAILURE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "failure-tempest-1",
    "messages": [
      {"role": "system", "content": "You are a careful assistant."},
      {"role": "user", "content": "Is 2^61 - 1 prime? Show the check."}
    ]
  }'

The response is a standard chat.completion object, with a tempest object alongside usage that tells you what the pipeline did.

4. Python

pip install openai

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://api.failure.fail/v1",
    api_key=os.environ["FAILURE_API_KEY"],
)

resp = client.chat.completions.create(
    model="failure-tempest-1",
    messages=[{"role": "user", "content": "Is 2^61 - 1 prime? Show the check."}],
)
print(resp.choices[0].message.content)
print(resp.usage)

5. TypeScript

npm install openai

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.failure.fail/v1",
  apiKey: process.env.FAILURE_API_KEY,
});

const resp = await client.chat.completions.create({
  model: "failure-tempest-1",
  messages: [{ role: "user", content: "Is 2^61 - 1 prime? Show the check." }],
});
console.log(resp.choices[0].message.content);
console.log(resp.usage);

6. Stream it

Fused answers take longer than a single model on hard prompts. Stream so your users see the working as it happens: it arrives as reasoning_content deltas before the answer arrives as content.

stream = client.chat.completions.create(
    model="failure-tempest-1",
    messages=[{"role": "user", "content": "Plan a zero-downtime Postgres major upgrade."}],
    stream=True,
)
for chunk in stream:
    if chunk.choices:
        delta = chunk.choices[0].delta
        working = getattr(delta, "reasoning_content", None)
        if working:
            print(working, end="", flush=True)
        if delta.content:
            print(delta.content, end="", flush=True)
    if chunk.usage:
        print("\n", chunk.usage)

More in streaming.

Next steps

• Choose the right model id — models.
• Cut cost on repeated context — caching.
• Handle the 402 and 503 cleanly — errors.