Hy Image 3.5 Preview use_search_tool is off by default: what Sume has

Hy Image 3.5 Preview has an external search switch called use_search_tool, and Tencent says it is disabled by default. Hy Image 3.5 Preview is not listed in the Sume image catalog (read 2026-10-09). The Sume request has no search or grounding flag for any image row, so a search-aware image has to come from you pasting the facts into the prompt.

How does the switch work at Tencent?

The field is an object, not a boolean, and its structure is fixed.

What would a request look like?

This is the shape from Tencent's page, with the search switch on. It goes to Tencent's TokenHub endpoint, not to Sume.

{
  "model": "hy-image-v3.5-preview",
  "session": "demo-1",
  "use_search_tool": {
    "value": true
  },
  "messages": [
    {
      "role": "user",
      "content": [
        {
          "type": "text",
          "text": "A poster of this week's weather in Seoul"
        }
      ]
    }
  ]
}

What can I do on Sume instead?

Nothing equivalent exists in the request. Nano Banana 2.1 has search grounding at Google (5,000 free requests a month, then $14 per 1,000, read 2026-10-09), but the Sume image request has no flag to turn it on. Put the facts in the prompt: dates, names and numbers you looked up yourself, and ask the model to render exactly that text.

Listed alternatives for text-heavy posters are Nano Banana 2.1, GPT Image 2.5 and Ideogram 4.5 in the image models docs.

Is this worth waiting for?

Only if live facts in the picture are the product. For fixed copy, a looked-up prompt is just as good and costs one call.

If your poster needs live data, fetch it first in your own code, check the numbers, and then pass them in the prompt as quoted text. That keeps a human-checkable source for every figure on the image, which a hidden search step does not give you. The cost is one extra step before the image call, and no extra fee on Sume.