Models Reference

Fluxpool aggregates a curated set of image and video models behind one OpenAI-compatible API. The catalog changes over time — the authoritative list of what you can call today is served by the API itself.

List available models

Call GET /v1/models to get the live catalog for your account. Every returned id is a valid value for the model parameter on the generation endpoints.

curl https://api.fluxpool.ai/v1/models \
  -H "Authorization: Bearer fp_live_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"

from openai import OpenAI

client = OpenAI(
    base_url="https://api.fluxpool.ai/v1",
    api_key="fp_live_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"
)

for m in client.models.list().data:
    print(m.id)

A trimmed sample response:

{
  "object": "list",
  "data": [
    { "id": "flux-2-pro", "object": "model" },
    { "id": "seedance-v2", "object": "model" },
    { "id": "qwen-image-max", "object": "model" }
    // ... more entries
  ]
}

Using a model ID

Pass any ID returned by /v1/models as the model parameter on the relevant endpoint. Image models go to /v1/images/generations, video models to /v1/videos/generations. Or use the auto-routing /v1/generations endpoint.

# Image example
img = client.images.generate(
    model="flux-2-pro",
    prompt="A calm mountain lake at sunrise, ultra realistic",
    size="1024x1024"
)
print(img.data[0].url)  # S3 presigned URL, expires in 1 hour

Verified example models

These two model IDs are always safe to reach for when you just want to test the API end-to-end. Both are hit by our production smoke tests.

Every other model below is also live. GET /v1/models is the authoritative list for your account.

All models

The 16 models on Fluxpool today. Pass the ID as model. Capabilities and live prices are on the Models and Pricing pages.

Model lifecycle

Model IDs are stable for the life of the model. When a model is retired, it stops appearing in GET /v1/models and returns a 400 if called. We announce retirements in the Changelog ahead of the sunset date.

To make your integration resilient: fetch /v1/models at startup (or cache it briefly) rather than hard-coding a list in your app.