What Is BAGEL Image?
BAGEL is ByteDance Seed’s unified multimodal foundation model with 7B active parameters. This page exposes its exact text-to-image path for detailed square compositions while keeping the optional paid thought-token pass disabled.
What Is BAGEL Image Best For?
- Dense fantasy scenes
- Unified-model experimentation
- Square concept art
- Complex compositional prompts
How to Use BAGEL Image
- Write a detailed scene with clear spatial relationships.
- Use square composition language.
- Refine with one output first.
- Generate a small final batch after the prompt stabilises.
BAGEL Image Settings Explained
Aspect Ratio
The exact generation operation returns a 1024×1024 image, so only square is offered.
Seed
Reuse a seed when tightening a complex prompt.
The panel above shows only the parameters BAGEL Image actually supports. Switch versions and unsupported controls disappear rather than sitting there doing nothing.
BAGEL Image Prompt Examples
World-building
Ancient observatory floating above storm clouds, crystalline arches, tiny lantern-bearing figures, two moons, cinematic depth
Object study
Intricate brass astrolabe on black velvet, museum lighting, engraved markings, centred square composition
BAGEL Image Test Results
Testing in progress
Every model runs the same eight-category prompt set — portrait, product, poster text, anime character, interior, multi-reference, image editing and character consistency — recording generation time, success rate, text accuracy and identity retention. That run has not finished for BAGEL Image, so this page shows no scores. We do not publish numbers we have not measured.
BAGEL Image Limitations
- Square 1K output on this endpoint
- No reference editing in this first route
- Thought-token generation is disabled to keep price deterministic
BAGEL Image Pricing and Credits
Credits scale with resolution and image count. The Generate button always shows what this run costs.
1K
8 credits / image
Failed generations cost 0 credits.
Commercial Use and Image Rights
BAGEL is released under Apache 2.0; generated use remains subject to applicable law and site terms.
- Model used: BAGEL Image
- Model developer: ByteDance Seed
- The inference route is managed internally and may change without changing the selected model
- Uploads are sent only to the configured inference service needed to complete the generation
- You can delete any generation from your history at any time
Frequently Asked Questions
What is thought mode?+
The upstream option spends extra tokens and raises price by 20%; it is disabled until the checkout can quote it explicitly.
Can BAGEL understand images too?+
The broader model can, but this page intentionally calls only the exact text-to-image operation.
Release History and Sources
May 20, 2025 · Model and research released · open weights
This date applies to the event shown. Later API access, weights and availability on All Image AI may have different dates.
Explore this event on the image model timeline →
Related Models and Tools
Seedream 5 Pro
ByteDance
A production model for coherent scenes, accurate edits and reference-heavy creative work.
Seedream 5 Lite
ByteDance
A lower-cost Seedream 5 tier for fast generation, reference-led edits and native 4K output.
Seedream 4.5
ByteDance
High-fidelity multi-image editing with stronger typography than Seedream 4.
Seedream 4
ByteDance
Native 4K, bilingual text, built for producing sets rather than single images.