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.

ByteDance's Seedream 5.0 Lite page describes two headline features: online search and thinking. On Sume, bytedance/seedream-5.0-lite is a plain image row. It lists 9 aspect ratios, bills about $0.04375 per image, and requires provider.options to be empty, so neither feature has a switch in the Sume request. You send a prompt, optional references and a ratio, and you receive an image.
That is a real boundary to know before you design around the vendor page. If the web search step is what you need, put it before the image call, in your own code or in an agent, and feed the result into the prompt.
Vendor page vs Sume row
| Topic | Vendor page | Sume row |
|---|---|---|
| Online search | Described as a feature | No request field |
| Thinking | Described as a feature | No request field |
| Aspect ratios | Not compared here | 9 listed |
| Price | Not compared here | About $0.04375 per image |
| Provider options | Not applicable | Must be empty |
How to get the effect without the flag
- Do the research first: look up the facts the image needs (a landmark, a product spec, a current logo shape) and write them into the prompt.
- Plan the layout in text first, then paste the plan into the image prompt as plain sentences.
- Pass one or two reference images when the subject must be accurate; the row takes up to 10.
- Check the output against the facts you looked up, and rerun with a tighter prompt if it misses.
Price check
At about $0.04375 per image, 100 images cost $4.38 and 1,000 cost $43.75. Seedream 4.0 on Sume is cheaper at $0.0325 and Seedream 4.5 costs $0.05. If you do not need what 5.0 Lite adds, test 4.0 first on the same prompts, since the price gap is about a third.
For ratios, 5.0 Lite and 4.5 each list 9 and 4.0 lists 10, so a ratio that only 4.0 offers is a reason to stay on 4.0.
A prompt worth testing
Try one prompt that needs a current fact, such as a poster for a named public event with the date in the headline. Write the date into the prompt yourself, since the row will not look it up. Then try the same prompt without the fact and compare. The difference shows how much of the vendor's search feature you are replacing with your own research step.
Keep the research text short. Two sentences of facts help; a pasted article makes the prompt harder to follow and does not make the image more accurate.
What Sume does not do
Sume does not run a search step for you and does not expose a thinking switch on this row. Sume also does not confirm that the vendor page's features are enabled upstream for your request, as it only sends the fields in the catalog. If a later catalog version adds a descriptor, the endpoints route will list it.
Sources
Related posts
More in Models
- 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.
- sume/music-auto or pinned lyria-3.5: what changes in the job
Music Router takes sume/music-auto (Lyria 3.5 today), lyria-3.5 or lyria-3-pro. What job.model and routed_model echo back, and why the price stays the same.
Written by Sume