Skip to content
AxAxiom

Control plane, not a playground

The API is real. Generation is not switched on.

Axiom is a working name for an account, tenant, and job control plane aimed at small developer teams. Catalog models stay candidates until a provider, a license review, safety controls, benchmarks, and billing are actually configured. This deployment will not invent an image, a price, or a benchmark.

Inference
Disabled
Card billing
Not configured
Public models available
0
Database
Connected

What works in this build

  • Accounts with Argon2id passwords and session cookies. API keys require a verified email. Mail is sent only when SMTP is configured.
  • Organizations, revocable API keys shown once, and tenant-scoped jobs.
  • A prepaid credit ledger in integer USD cents: reserve, settle once, release.
  • Async jobs with a required idempotency key, cancellation, and a Postgres-backed worker.
  • Private uploads, signed download URLs, and signed webhook deliveries.

What is intentionally off

  • No GPU, no hosted inference, and no call to a generation vendor.
  • FLUX.1 schnell and Wan2.2 TI2V-5B are registry candidates, not live models.
  • No card charges, checkout, or published per-output prices.
  • No sample customers, fake progress, or benchmark scores.

The request you can send today

Returns model_not_available

A verified API key can call POST /v1/jobs. Candidate models fail before a job row or credit reservation is created. The response below is the one this build returns for the documented example.

curl -sS -X POST https://axiomgen.dev/v1/jobs \
  -H "Authorization: Bearer $API_KEY" \
  -H "Idempotency-Key: 3f1c0c4e-8a2b-4d7e-9c11-6b0a2f5d1e77" \
  -H "Content-Type: application/json" \
  -d '{"model":"flux-1-schnell","version":"candidate-1","input":{"prompt":"A ceramic cup on a walnut table, morning light","width":1024,"height":1024}}'
{
  "error": {
    "code": "model_not_available",
    "message": "flux-1-schnell@candidate-1 is registered as \"candidate\" and is not available for generation. No provider was called.",
    "request_id": "generated-per-request"
  }
}