ByteDance’s Seed team released Seedream 5.5, a new image creation model, on 9 November, according to its official blog post. The post says the model supports up to 100 rounds of continuous editing on the same image while keeping the areas that are not being edited unchanged at the pixel level.
The model is rolling out on Jimeng, Doubao, Doubao Work and Volcano Engine Ark, ByteDance says. Tech outlet Leiphone reports that a 2K square-image edit costs about 0.28 yuan through the API, less than half the Seedream 5.0 Pro price; ByteDance’s own post gives no price.
Business Pill · Tokens Explained
An evergreen Business Engineer explainer on how AI models are metered and priced. It teaches the general idea only and does not describe Seedream 5.5.
The key insight: ByteDance released Seedream 5.5 on 9 November with up to 100 rounds of continuous editing and native transparent output, according to its Seed blog, and Leiphone reports API edits cost about 0.28 yuan, less than half the Seedream 5.0 Pro price.
The blog names three upgrades. The first is the multi-round editing: in our translation, it says Seedream 5.5 can support “up to 100 rounds of continuous creation on the same work”, with “pixel-level preservation of non-edited areas”.
The second is native transparent output. The post describes “native RGBA transparent asset generation” that “directly outputs transparent or semi-transparent images with an alpha channel”, which it says lets “generated elements enter later design workflows as standalone assets”.
The third is image realism and text. The post says lighting, materials and skin texture look more natural, and that the model renders dense small text and complex Chinese characters more accurately. ByteDance also says a Pro version of the 5.5 series will follow.
Release dates are from ByteDance’s own Seed and BytePlus blogs; the day counts between releases and the 341-day span are our arithmetic.
ByteDance’s Own Evaluation Numbers
All of the benchmark figures come from ByteDance’s own tests on its MagicArena platform, scored by expert raters across six scenarios: e-commerce, marketing, film and TV, games, office, and fashion design. Against Seedream 5.0 Pro, the post says scores and usable rates rose in all six, and the usable rate in film and TV and in office was more than twice the previous version’s.
On text, ByteDance reports a severe text-breakage rate of 0.4% and a minor text-breakage rate of 1.7%, which it defines as the share of generated images with wrong characters, missing characters or broken structure. In realism for image-to-image tasks, it reports a Good/Same/Bad preference score of +55.62% in a comparison with the previous generation, and says the model leads GPT Image 2.5 Sunburst and Nano Banana 2.1 on text structure and text-to-image realism.
Price and Availability
Leiphone says Doubao, Doubao Work and Jimeng already carry Seedream 5.5 and that the API will soon go live on Volcano Engine Ark. It reports that billing keeps the two-part input-image plus output-image structure and that text tokens are free.
At about 0.28 yuan for a 2K square-image edit and less than half the old price, Leiphone’s figure implies Seedream 5.0 Pro cost more than about 0.56 yuan for the same edit, by our arithmetic. A third-party reseller, Atlas Cloud, already lists a Seedream 5.5 edit endpoint; it says it provides access and does not own the model.
How Fast the Versions Are Coming
ByteDance’s own blogs date Seedream 4.5 to 3 December 2025, Seedream 5.0 Lite to 24 February 2026 and Seedream 5.0 Pro to 8 July 2026. That puts 124 days between 5.0 Pro and 5.5, after gaps of 83 and 134 days, by our arithmetic.
As we read the release, ByteDance pitches Seedream 5.5 at production work rather than one-off images: its own tests cover e-commerce, marketing, film and TV, games, office, and fashion design.
As we read the dates on ByteDance’s own blogs, Seedream 5.5 arrives 124 days after Seedream 5.0 Pro, the fourth release in 341 days, by our arithmetic.
ByteDance Seed release blog, our translation
“compared with Seedream 5.0 Pro, Seedream 5.5 improved both the generation score and the usable rate in all six scenarios; in film and TV and in office, the usable rate reached more than twice that of the previous version.”
Three Implications
FOR DESIGNERS ByteDance says the model directly outputs transparent or semi-transparent images with an alpha channel.
FOR API BUYERS Leiphone reports text tokens are free and a 2K square-image edit costs about 0.28 yuan.
FOR RIVALS ByteDance says Seedream 5.5 leads GPT Image 2.5 Sunburst and Nano Banana 2.1 on text structure and text-to-image realism, in its own tests.
The framework says: “Inference price per equivalent unit of work has collapsed by a similar order of magnitude over the same window.”
As we read it, Leiphone’s report fits that pattern: it says a Seedream 5.5 edit costs less than half what the same edit cost on Seedream 5.0 Pro.
What Is Not Established
The evaluation figures are ByteDance’s own and have not been independently reproduced. ByteDance’s post does not give a price; the 0.28-yuan figure is Leiphone’s. The API launch date on Volcano Engine Ark, international availability through BytePlus, and the Pro version’s timing are not stated.
This Week In The AI Supercycle
The Great Convergence of Capital, Power, and Geopolitics
This week’s issue: for two years the AI build-out was priced as a supply story, and last week as a credit story. This week it was priced as a permission story, and the answer arrived as a cost, not a ban.
ByteDance released Seedream 5.5 on 9 November with up to 100 rounds of stable editing and native transparent output, 124 days after Seedream 5.0 Pro by our arithmetic, and Leiphone reports its API edits cost less than half the previous version’s.
A note on sourcing. Seedream 5.5’s features and evaluation figures are from ByteDance Seed’s 9 November 2026 release blog, in Chinese; quotes are our translation. The price and API timing are from Leiphone’s 9 November report. Earlier release dates are from ByteDance’s Seed and BytePlus blogs. Day counts and the implied earlier price are our arithmetic.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
