Quickstart

Enable AI for your project, link a USD billing account and create a credential with ai:read and ai:generate. Operator activation and reviewed prices are required. An empty catalog is not a promise that arbitrary provider model names will work.

1. Find a model

export FEEEF_CLOUD_KEY='your-project-key'
curl "https://apis.cloud.feeef.org/v1/models" \
  -H "Authorization: Bearer $FEEEF_CLOUD_KEY"

Pick an exact data[].id whose endpoints contains chat/completions. Review its constraints and price version. Do not substitute a provider's unlisted ID.

2. Generate text

import OpenAI from 'openai'
import { randomUUID } from 'node:crypto'
const ai = new OpenAI({
  apiKey: process.env.FEEEF_CLOUD_KEY,
  baseURL: 'https://apis.cloud.feeef.org/v1',
  maxRetries: 0,
})
const requestKey = randomUUID() // Save alongside your job before submission.
const result = await ai.chat.completions.create({
  model: 'MODEL_ID_FROM_CATALOG',
  messages: [{ role: 'user', content: 'Explain what a webhook does.' }],
  max_completion_tokens: 256,
}, { headers: { 'Idempotency-Key': requestKey } })
console.log(result.choices[0]?.message.content)

Adjust the output-limit field to the selected model's constraints. If you retry, reuse the same key and identical payload. A new key starts another billable request.

3. Inspect the charge

Keep the response's X-AI-Request-Id. Read /v1/ai/tasks/{id} with the same project and tenant context to inspect verified billing. New requests have no upfront hold; actual usage is charged after verification. Try the playground to view model capabilities, results and billing together.