Estimate a Sume image batch cost from endpoint pricing lines in Python

The Image API docs state that endpoint pricing lines are the amount charged to your wallet, with margin already applied, so cost_usd times n is what you pay. For a batch, fetch the model's endpoint, take the output_image line and multiply by the number of images. Where a model prices by token or by quality, the line may not be a flat per-image number, so read what it lists.

What the docs say

From the Image API docs.

The estimator

It handles only flat per-image lines and says so otherwise.

import os, requests

def estimate(model, n_total):
    r = requests.get(
        f"https://api.sume.com/v1/images/models/{model}/endpoints",
        headers={"Authorization": f"Bearer {os.environ['SUME_API_KEY']}"},
        timeout=30,
    )
    lines = r.json()["endpoints"][0]["pricing"]
    flat = [p for p in lines if p["billable"] == "output_image" and p["unit"] == "image"]
    if not flat:
        raise ValueError(f"no flat per-image line: {lines}")
    return round(flat[0]["cost_usd"] * n_total, 4)

print(estimate("bytedance-seed/seedream-4.5", 200))

Caveats

The catalog changes, so read prices at run time. Quality and size can change the charge on some models; for ChatGPT Image 2.5 the docs say auto quality reserves the max amount, so pin quality before you estimate. Confirm after the run with GET /v1/usage.