Is Hy Image 3.5 Preview on Sume? Check the models list in code

No: Sume does not list Hy Image 3.5 Preview in the image catalog this post was checked against, so there is no Sume model id to send for it. The way to be sure on the day you read this is to ask the catalog: GET /v1/images/models returns every image id Sume serves, and a 12-line script can test for any name.

What Tencent released

LLM Gateway's timeline records that Tencent Cloud released Hy Image 3.5 Preview on October 5, 2026, with the model id hy-image-v3.5-preview. OpenRouter lists it as tencent/hy-image-v3.5-preview, a unified text-to-image, image-to-image and multi-turn editing model that takes up to 20 reference images and produces output up to 4K.

• OpenRouter resolutions: 1K, 1.5K, 2K, 4K.
• OpenRouter limit: 1 image per request.
• Sume has a different catalog, so these limits do not carry over to any Sume row.

Test the catalog instead of guessing

The Image API docs say the models endpoint lists the models and their capabilities. This script prints whether any id contains a name fragment, and lists the edit-capable ids when it finds nothing.

import json, os, urllib.request

req = urllib.request.Request(
    "https://api.sume.com/v1/images/models",
    headers={"Authorization": "Bearer " + os.environ["SUME_API_KEY"]},
)
with urllib.request.urlopen(req, timeout=30) as r:
    models = json.load(r)["data"]

needle = "hy-image"
hits = [m["id"] for m in models if needle in m["id"]]
print("found:" if hits else "not listed:", hits or needle)
if not hits:
    for m in models:
        if "image" in m["architecture"]["input_modalities"]:
            print("edit-capable:", m["id"])

Reading the result

The script prints found: with matching ids or not listed: with the needle. Change needle to check ideogram, nano-banana or gpt-image-2.5 the same way. Each record also carries supported_parameters, so you can extend the loop to test whether a row accepts input_references or lists the aspect ratio you need before you submit a paid call.

The per-model endpoints route, GET /v1/images/models/{model_id}/endpoints, gives the definitive parameters and the pricing lines for one model. Sume publishes a single sume endpoint per model in v1, so there is no provider choice to make once you have the id.

If it is not listed

Pick the closest row by capability instead of by name. For many references, ChatGPT Image 2.5 on Sume accepts up to 16 image references and an optional mask_url. Ideogram 4.5 edits the first image and takes up to 4 more as references, 5 in total.

Do not send a Tencent or OpenRouter id to Sume and expect it to work; the id has to come from the Sume catalog. The only automatic choice is model: "sume/auto", which never discloses which family ran, so it is not a way to reach a specific outside model.

One more habit helps: store the full model list response in your repository's test fixtures once a week and diff it. A new row, a retired row or a changed ratio list then shows up as a code review comment instead of a production error.

If Sume lists Hy Image later, the models endpoint will show it and this script will print it. Until then, treat any comparison with Hy Image as a comparison against another vendor's page, not against something you can call here. Tencent describes the model as a preview, and preview models can change; that is one more reason to read the live catalog on the day you build.

What the OpenRouter page does and does not say

The OpenRouter page describes capabilities, not a Sume offering: reference-image count, resolution tiers, ratios and a one-image-per-request limit. It shows a sample usage charge in USD for a token count but no rate card, so this post gives no Hy Image price. Tencent's own price sheet was not fetched for this post, and any figure quoted elsewhere should be traced to the vendor page before you plan a budget around it.