Seedream 4.0
Image Available Full comparison ↗ bytedance/seedream-4-0 · by ByteDance Seed
· Diffusion transformer (DiT)
Pricing — 1 offering(s)
Default
- $0.03 / image
Current
2026-07-19 → present
Flat rate regardless of output resolution (1K-4K, up to 4096x4096). API model id: seedream-4-0-250828…
Flat rate regardless of output resolution (1K-4K, up to 4096x4096). API model id: seedream-4-0-250828. effective_from is the date this entry was sourced, not a confirmed price-change date.
Showing the active price and any recorded history. Full pricing history is available via the paid API — see API docs.
See how Seedream 4.0 fits into a cost-aware routing setup
See how →Capability profile
How Seedream 4.0 rates across core capability dimensions, with the task-level evidence behind each rating.
ByteDance's own blog claims it 'broke through bottlenecks of previous generative models in text processing,' with correct rendering of formulas, tables, and charts specifically called out.
4K generation, up from Seedream 3.0's 2K ceiling, per ByteDance's own blog.
Reported >10x faster than Seedream 3.0 via quantization, speculative decoding, and adversarial distillation — a vendor claim, not independently timed here.
Supports up to a dozen reference images for organic multi-image composition/fusion, per ByteDance's own blog — a real, specific capability claim, not generic marketing.
Operator guidance
Route here for fast generation with strong text/formula/table rendering and multi-image composition. Compare against Gemini 3.1 Flash Image or Gemini 3 Pro Image (also strong on text rendering) — no independently verified head-to-head exists between any of these in this registry.
Use cases
- In-image text, formulas, tables, and chart rendering
- Multi-image composition/fusion from up to a dozen reference images
- Fast, high-volume generation and editing in one unified model (background replacement, portrait retouching)
- Structured reasoning-adjacent generation — ByteDance's own blog specifically mentions puzzles, crosswords, and comic-strip continuation
Limitations
- Prompt adherence, photorealism, and artistic range are unrated — no independent or vendor-specific claim distinct from generic marketing found
- ByteDance's own blog cites benchmark wins (highest Elo in single-image editing, leading text-to-image on "MagicArena," beating Gemini 2.5 Flash) on ByteDance's own MagicBench evaluation — a vendor-run benchmark, not one of this registry's tracked independent benchmarks (GenEval/T2I-CompBench/etc.), so deliberately not recorded in the structured benchmarks field; flagged here as a vendor claim, not verified