Nagent AI
AI Models/ByteDance AI
ByteDance

ByteDance AI

Image GenerationVideo Generation

Seedream 5.0 (image) and Seedance 2.0 (video) — commercial-batch generation at scale.

ByteDance's generative stack — Seedream 5.0 for image and Seedance 2.0 for video — has emerged as the industry leader for e-commerce and product photography. Seedream is optimised for batch consistency (dozens of variations of the same product across lifestyles or environments). Seedance 2.0 features a unique '9+3+3' reference system (9 images + 3 videos + 3 audio clips as input for a single generation), the highest level of director control on the market. Note: Seedance 2.0 is in active legal disputes with Hollywood studios over training data, which may impact long-term Western-market availability.

Access on Nagent

Models available3
Modalities2
Available on Nagent

Models

Seedance 2.0

Latest

Highest director control via 9+3+3 reference system

Input Price

$0.10 – $0.80 / sec

Input Types

Text prompt, Up to 9 images + 3 videos + 3 audio clips

Output Types

Video (1080p, up to 15s)

Seedance 1.5 Pro

Stable previous-gen tier

Input Types

Text prompt, Image

Output Types

Video (up to 60s)

Seedream 5.0

Reliable

Best for product batch consistency in e-commerce

Input Price

~$40 per 1K images

Input Types

Text prompt, Reference image

Output Types

Images

What You Can Build

Use Cases

E-Commerce Product Batches

Generate dozens of variations of a single product in different lifestyle contexts while keeping the product itself perfectly consistent.

Director-Controlled Video

Use Seedance's 9+3+3 reference inputs to produce video that closely matches a specific creative brief — the highest level of control we've seen.

TikTok-Native Format

ByteDance models understand short-form social content patterns natively — output is platform-ready without post-processing.

Localised Content at Scale

Generate regionally adapted creative for APAC and global markets in the same workflow.

Platform Advantage

Why use ByteDance AI on Nagent?

Nagent adds enterprise orchestration, observability, and workflow automation on top of ByteDance AI's raw model capabilities.

Seedream batch consistency is wired into our Ad Genie agent for e-commerce campaigns

Seedance 9+3+3 inputs map cleanly to our multi-asset workflow primitives

Provider-failover ready: if Seedance is unavailable in your region, KARMIC auto-routes to Veo / Kling

Highest throughput in our stack for short-clip generation

Getting Started

How to access ByteDance AI on Nagent

01

Open Agent Studio

Navigate to Agent Studio in your Nagent workspace.

02

Select ByteDance AI

Choose Seedance or Seedream under Video / Image Generation.

03

Connect Social Tools

Wire output to TikTok Business API, Meta Ads Manager, or your scheduler via Nagent connectors.

FAQs

Common questions about ByteDance AI

Real buyer and developer questions, answered. Click any item to expand.

What is the "9+3+3" reference system?

You can input up to 9 images, 3 videos, and 3 audio clips as references for a single video generation. It's the highest level of director control on the market — closest thing to a creative brief that the model actually respects.

Will the Hollywood lawsuit affect my access to Seedance?

Possibly in Western markets long-term. Nagent's failover routing to Veo / Kling is automatic if Seedance becomes unavailable in your region — your agents continue running without code changes.

When should I pick Seedream over Ideogram for image generation?

Seedream for product batches — same product across many lifestyles or environments while keeping the product itself perfectly consistent. Ideogram for typography-heavy design work. They're complementary, not competing.

Is ByteDance's data sent to China?

Inference runs on whichever endpoint you configure. For Western deployments use the US/EU endpoints; for Asia-Pacific the regional endpoints. Confirm specific data flows with your compliance team before deploying regulated workloads.

Ready to use ByteDance AI inside your agents?

Get started in minutes — no API key management required.

All model providers