Seedream 5.0 Lite 'online search': not a Sume image parameter
ByteDance says Seedream 5.0 Lite can use online search. Sume's seedream-5-lite row has no search field. What the row offers: ratios, 10 references, $0.04375.
ByteDance's Seedream 5.0 Lite page describes a model with deep thinking and online search, so news and trending topics can be visualized. Sume's bytedance-seed/seedream-5-lite row does not expose a search or thinking switch. The request schema has no such field, and a field the catalog does not list returns 400 unsupported_parameter.
If you want a picture of something that happened today, you bring the facts yourself, in the prompt or as a reference image. The row gives you the image model, not a retrieval step.
What the Sume row does list
| Field | Value |
|---|---|
| Billed per image | about $0.04375 (list $0.035 x 1.25) |
| Aspect ratios | 1:1, 16:9, 9:16, 4:3, 3:4, 4:5, 5:4, 3:2, 2:3 |
| n | 1 to 4 |
| input_references | up to 10 |
| output_format | png, jpeg, webp |
| Custom pixels | image_size {width,height} |
A manual way to get search-grounded art
Do the retrieval yourself and pass the result in. For a news-style graphic:
- Collect the facts: names, numbers, the date. Write them into the prompt as plain text.
- If a logo or product photo matters, upload it to a public HTTPS URL and send it in
input_references. - Ask for text sparingly. Rendering a long headline is the part most likely to need a retry; test it first at a low-cost row.
- Generate with
n: 2to get a spare, then keep one.
Prompts that work without search
Put the date and the specific details in text: 'a clean poster for the October 31 event, orange and black, big numeral 31 at center'. Seedream-style models follow layout instructions well, and ByteDance's page says every detail aligns with your instructions, so lean on specifics rather than on a hope that the model knows today's news.
Use aspect_ratio to match the channel, 4:5 for a feed post or 9:16 for a story, and keep n at 2 while you iterate.
Why this difference matters for budgeting
A model with built-in retrieval may take longer or cost more per call, and that is the vendor's decision. On Sume the price is fixed per image at the catalog rate, so a news-graphic batch is easy to price: 40 images are about $1.75. What you do not get is an automatic fact check. If the image shows a statistic, someone on your side verifies it.
If a slow generation falls past the 30-second wait, the route returns 202 with a job envelope, and you read the result from the job endpoints. Branch on the status code, not on the body shape.
What Sume does not do
Sume does not fetch web pages, search the web or ground an image on Sume's side for the Seedream row. It also does not tell you what the upstream model did internally. If grounding is the point, say so in your tooling and add the retrieval step to your own pipeline.
Sources
Related posts
More in Models
- Seedream 5.0 Lite search and thinking: what Sume's API passes
ByteDance describes online search and thinking for Seedream 5.0 Lite. Sume's row lists 9 ratios at about $0.04375 and empty provider.options. What that means.
- Seedream 5 Lite or 4.5: both 5 cents on Sume, and v4 is 4 cents
Seedream 5.0 Lite and 4.5 both bill 5 cents on Sume; Seedream 4.0 bills 4. Their listed ratios match and all three edit. What the catalog does and does not say.
- Sonic 3.6 vs 3.5 vs 3 on Sume TTS: same price, so pin the id
All Sonic ids in Sume's TTS router list 38 micro-dollars per character and a 20,000-character cap. sonic-latest is an alias for 3.6. What to pin and why.
- Strands Decider 2B scores 167 of 231: not a spend guard
The Strands Decider 2B card reports 167 of 231 correct on JevBench public (72.3%). Fine for routing, not payment control. Why Sume's spend cap decides.
Written by Sume