ByteDance Launches Seedream 5.5: 100 Rounds of Stable Edits, Native Transparent Output
ByteDance’s Seed team released the image model Seedream 5.5 on Nov 9: a single image can go through up to 100 rounds of local edits while keeping composition and lighting stable, with native transparent-background output; a 2K edit costs about 0.28 yuan, and the model is live in Dreamina, CapCut, Doubao and other products.

What happened
Multi-round editing is the core of this upgrade: Seedream 5.5 supports up to 100 rounds of continuous local edits on a single image, with ByteDance saying pixels in unedited areas stay stable and composition, perspective and lighting hold across many revisions; one edit can adjust furniture, lighting, hairstyles, outfits and other local details.
Transparent output and text rendering improve together: the model supports native RGBA output, producing transparent-background images and layered assets directly; ByteDance reports a severe text-corruption rate of 0.4% and a minor rate of 1.7%, covering Chinese and English including dense small text; its GSB preference score for image-to-image realism is 55.62% higher than Seedream 5.0 Pro, a vendor benchmark figure.
Channels and pricing: the model is live in Dreamina and CapCut, with Doubao, BytePlus and Volcano Engine Ark also connected; a 2K edit costs about 0.28 yuan; the rollout is gradual and not yet open to all accounts and regions; a Seedream 5.5 Pro version follows later, with no date given.
Key facts
- Released by: ByteDance’s Seed team, Nov 9, 2026
- Multi-round editing: up to 100 rounds of continuous local edits on one image, with unedited pixels staying stable
- Transparent output: native RGBA, producing transparent backgrounds and layered assets directly
- Text rendering: vendor-reported severe corruption rate 0.4% and minor rate 1.7%; GSB preference score 55.62% above 5.0 Pro (vendor figure)
- Channels and pricing: Dreamina, CapCut, Doubao, BytePlus and Volcano Engine Ark; about 0.28 yuan per 2K edit; gradual rollout; Pro version to follow
Why it matters
Image models are shifting from one-shot generation to multi-round production editing, and transparent output with low text-error rates maps directly onto design and e-commerce workflows.
CompaniesByteDance
Grab todayOpenAI Codex for Students: $100 in Free CreditsHIMIAO Picks →Follow HIMIAO Horizon
No sign-up · Unsubscribe any timeNotifications
A ping on your phone or computer when the brief is out. No email needed.
Comments
Related
- Oct 21ByteDance, Alibaba and Tencent Place Over $35 Billion in Nvidia Chip Orders, Buying Up China-Compliant RTX PRO 5500
- Nov 12MiniMax Launches H3-Image: 4K Images in About 10 Seconds, Open Weights With a Restricted License
- Oct 16Black Forest Labs Releases FLUX 3 Image Klein: 5.6B Open-Weights Image Model Runs on 8GB GPUs
- Oct 15HKUST(GZ)'s Open-Source Supervisor-Skills Tops GitHub Daily Trending: Distilling a Decade of PhD Supervision into AI Skills
Today
Nov 12 Thursday- Brief
- GitHubGitHub Weekly Vol.5: 6 Projects Worth a Look This Week
Tested or read — weeklyAll →
- Quiz
- Call
Latest news
All →- Nov 12Kimi K3.1 Pro vs Qwen3.8-Max: Same $2 Input, $4 Apart on Output
- Nov 12NEAR AI Breached via SGLang Flaw, Exposing About 100 Users and 18 Agents
- Nov 12PLA-Linked Chinese University Bought Nvidia H200 Chips Despite US Blacklist
- Nov 12Ten AI Firms Sign Paris Call on AI Safety, Pledging Shutdown Mechanisms Without Penalties
- Nov 12OpenAI Acquires AI Travel Agent Odessia; Founder Joins, Service Shut Down
- Nov 12Meta's Infrastructure Chief Janardhan to Step Down in Early 2027
™