Models-Endpunkt

Die Models-API ermöglicht Ihnen:

• verfügbare Modell-IDs und Einstellungen zu entdecken
• eine Modellausführung pro Anfrage durchzuführen

Basispfad:

• https://armox.ai/api/v1/models

GET /api/v1/models

Gibt alle verfügbaren Modelle und ihre Konfigurations-Metadaten zurück.

Anfrage

curl -X GET "https://armox.ai/api/v1/models" \
  -H "Authorization: Bearer sk_live_your_key_here"

Antwort

{
  "object": "list",
  "data": [
    {
      "id": "google/nano-banana",
      "name": "Nano Banana",
      "modality": "image",
      "cost": 70,
      "inputTypes": ["text", "image"],
      "promptRequired": true,
      "outputType": "image",
      "description": "Google image generation and editing model",
      "settings": [
        {
          "key": "aspect_ratio",
          "label": "Aspect Ratio",
          "inputType": "select",
          "defaultValue": "1:1"
        }
      ]
    }
  ]
}

POST /api/v1/models/run

Führt eine Modellgenerierung aus.

Request Body

{
  "model": "google/nano-banana",
  "input": {
    "prompt": "string",
    "image_input": ["https://example.com/image.jpg"]
  },
  "webhook_url": "https://your-app.com/webhooks/armox"
}

Felder:

• model (erforderlich): Modell-ID aus GET /api/v1/models
• input (optionales Objekt): modellspezifische Eingaben und Einstellungen
• webhook_url (optional): Callback-URL für Abschluss-Events

Synchrones vs. asynchrones Verhalten

POST /models/run kann entweder zurückgeben:

• 200 für synchrone Fertigstellung (typischerweise Bild-/Text-/Tool-Ausgaben)
• 202 für asynchrone Verarbeitung (Video-/Audio-/3D-Ausgaben)

Beispiel: Bildmodell-Ausführung (synchron)

Anfrage

curl -X POST "https://armox.ai/api/v1/models/run" \
  -H "Authorization: Bearer sk_live_your_key_here" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "google/nano-banana",
    "input": {
      "prompt": "Turn this into a cinematic night street photo with neon signs and reflections.",
      "image_input": ["https://example.com/reference-image.jpg"],
      "aspect_ratio": "match_input_image",
      "output_format": "png"
    }
  }'

Typische 200-Antwort

{
  "id": "a1b2c3d4-e5f6-7890-ab12-34567890abcd",
  "object": "job",
  "status": "completed",
  "model": "google/nano-banana",
  "output": {
    "outputImage": "https://storage.example.com/public/api-outputs/..."
  },
  "error": null,
  "credits_charged": 70,
  "created_at": "2026-03-26T12:00:00.000Z",
  "completed_at": "2026-03-26T12:00:03.000Z"
}

Beispiel: Videomodell-Ausführung (asynchron)

Anfrage

curl -X POST "https://armox.ai/api/v1/models/run" \
  -H "Authorization: Bearer sk_live_your_key_here" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "xai/grok-video-image-to-video",
    "input": {
      "prompt": "Animate this image with smooth camera movement and subtle environmental motion.",
      "image_input": "https://example.com/reference-image.jpg",
      "duration": 4
    }
  }'

Typische 202-Antwort

{
  "id": "f1e2d3c4-b5a6-7890-cd12-34567890efab",
  "object": "job",
  "status": "processing",
  "model": "xai/grok-video-image-to-video",
  "credits_charged": 202,
  "created_at": "2026-03-26T12:10:00.000Z"
}

Anschließend pollen:

curl -X GET "https://armox.ai/api/v1/jobs/f1e2d3c4-b5a6-7890-cd12-34567890efab" \
  -H "Authorization: Bearer sk_live_your_key_here"

JavaScript-Beispiel

const response = await fetch("https://armox.ai/api/v1/models/run", {
  method: "POST",
  headers: {
    Authorization: "Bearer sk_live_your_key_here",
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    model: "google/nano-banana",
    input: {
      prompt: "High-end product photo on clean white background",
    },
  }),
});

const job = await response.json();
console.log(job);

Python-Beispiel

import requests

response = requests.post(
    "https://armox.ai/api/v1/models/run",
    headers={
        "Authorization": "Bearer sk_live_your_key_here",
        "Content-Type": "application/json",
    },
    json={
        "model": "google/nano-banana",
        "input": {"prompt": "High-end product photo on clean white background"},
    },
)

print(response.status_code)
print(response.json())

Fehler, die Sie behandeln sollten

• 400 ungültiger Payload oder ungültiges Modell
• 401 fehlender/ungültiger API-Schlüssel
• 402 unzureichende Credits
• 403 Zugriff durch Abonnement eingeschränkt
• 429 Rate Limit überschritten

Vollständige Details finden Sie unter: Fehler & Rate Limits

Verwandte Seiten

• Jobs-Endpunkt
• Webhooks
• API-Referenz Einführung