Announced June 23, 2026 at Volcano Engine's FORCE conference, Seedance 2.5 adds 30-second native single-shot video and up to 50 omni-modal reference inputs. It's in enterprise closed beta now, with a public launch targeted for early July 2026 — not yet on any public API. Until it ships, Seedance 2.0 (now with 4K) is live and ready to use.
Status: enterprise closed beta · Public launch targeted early July 2026 · Vendor-stated specs, independently unverified
Everything announced at the June 23 reveal — stated by ByteDance, not yet independently benchmarked.
A single continuous shot of up to 30 seconds — no stitching. This is the headline upgrade over Seedance 2.0's 5–12 second clips.
Guide a generation with as many as 50 mixed references — text, images, video, and audio — for far tighter creative control.
Currently limited to a global enterprise closed beta. There is no public API access on kie.ai, fal.ai, or anywhere else yet.
ByteDance is targeting an early-July public release. No pricing, benchmarks, or independent sample reels exist as of this writing.
The same June reveal upgraded Seedance 2.0 with native 4K output. It's available to use on this site right now — no waitlist, no enterprise gate.
The 4K upgrade announced alongside 2.5 belongs to Seedance 2.0 — and you can generate in 4K here today.
Synchronized dialogue, ambient sound, and effects are generated together with the video in a single pass.
Combine text, images, video, and audio references to direct your shot exactly the way you want it.
Start generating free in seconds — while Seedance 2.5 is still locked to enterprise beta.
Straight answers on the 2.5 release date, features, pricing, and how it compares to Seedance 2.0.
Be first to know when a public Seedance 2.5 endpoint goes live. No spam — just the launch.
4K output, native sound, and multimodal input — live now, no waitlist. Try it free.