Release tracker · Updated as ByteDance ships
Seedance Version History — From 1.0 to 2.5 (and the Seedance 3.0 release date question)
Every Seedance video model release from ByteDance on one timeline. Seedance 2.5 was officially launched by ByteDance Seed on July 31, 2026 — live today on Jimeng AI and Doubao Pro, with BytePlus ModelArk API access marked "coming soon." Seedance 2.0 (released February 12, 2026) remains the current callable API model. Seedance 3.0 has no announced release date. We track all of it here, with sources, and update the moment something changes.
By Jay Yang · · 8 min read
30-Second Verdict
- 1Seedance 2.5 was officially launched by ByteDance Seed on July 31, 2026 (announced June 23, 2026 at the Volcano Engine FORCE conference). It rolled out first on Jimeng AI and the Pro version of Doubao, with API access via BytePlus ModelArk marked "coming soon" — as of August 2, 2026, no 2.5 API endpoint is callable yet.
- 2Seedance 2.0 (released February 12, 2026) remains the current callable API model, available on Volcengine Ark and BytePlus ModelArk under `dreamina-seedance-2-0-260128` at roughly $0.14 per second of generated video. All Seedance API integrations in production today still run on 2.0.
- 3Seedance 3.0 has no announced release date. As of August 2, 2026, ByteDance has not published any roadmap, specifications, or timeline for a Seedance 3.0 release — any "Seedance 3.0 release date" you see quoted today is speculation, not reporting.
Timeline — every Seedance release
In chronological order. Each entry includes status, public release date (or rumor source), one-line summary, and the source we used.
Seedance 1.0
HistoricalJune 2025
First public-facing Seedance video model, distributed inside Doubao and Jimeng.
- 5-8 second clips, 720p maximum, no native audio
- Mandarin-first prompt understanding; English supported but weaker
- Available only via ByteDance consumer surfaces (no public API)
Seedance 1.5 Pro
HistoricalLate 2025
Bridge release with improved temporal coherence and longer clips, ahead of the 2.0 architecture overhaul.
- 12-second maximum duration
- Multi-reference image input (image-to-video baseline)
- Internal name "1.5 Pro" — Pro suffix retired with 2.0
Seedance 2.0
Current callable APIFebruary 12, 2026
The current callable Seedance API model. First Seedance release with native synchronized audio, 15-second clips, and a public API on Volcengine Ark and BytePlus ModelArk. Remains the only Seedance model that can be integrated via API as of August 2, 2026 while the 2.5 endpoint is pending.
- 4-15 second clips, up to 2K resolution (native 480p/720p)
- Native synchronized stereo audio — BGM, SFX, voiceover, lip-sync, generated in one pass
- @-reference multimodal input: up to 9 images + 3 videos + 3 audio clips per request
- Public-beta API on Volcengine Ark since April 2, 2026 — pay-per-token, ~$0.14/sec
- Model IDs: `dreamina-seedance-2-0-260128` (BytePlus), `doubao-seedance-2-0-260128` (Volcengine); `-fast` and `-mini` variants also live
- Distributed via Volcengine Ark and BytePlus ModelArk (developers), Doubao and Jimeng (Chinese consumers)
Seedance 2.1
HistoricalRumored May 2026 — superseded by 2.5 launch
A May 2026 Pandaily report cited ByteDance sources promising a ~20% quality lift over 2.0. The 2.1 label never became a shipping product and was superseded by the Seedance 2.5 announcement (Jun 23, 2026) and official launch (Jul 31, 2026).
- Never formally launched — no Volcengine Ark page, no consumer surface, no API access
- Reported ~20% quality lift target (Pandaily, May 2026)
- Superseded in ByteDance's public roadmap by Seedance 2.5
Seedance 2.5
Launched · API pendingAnnounced June 23, 2026 · Officially launched July 31, 2026 · API pending
The newest Seedance generation. Officially launched by ByteDance Seed on July 31, 2026 with the post "One-take Creation, Flexible Referencing: Introducing Seedance 2.5." Live on Jimeng AI and Doubao Pro today; BytePlus ModelArk API access remains "coming soon" as of August 2, 2026 — no callable 2.5 endpoint exists yet.
- Confirmed by ByteDance Seed launch post: 30-second high-quality audio-video clips in a single pass, with multi-round extension for longer sequences
- Confirmed reference input: up to 30 images + 10 videos + 10 audio clips per request (vs. 9 + 3 + 3 on 2.0)
- New reference types: clay-render, motion, and creative references, plus timestamp-level editing / second-level control
- Consumer rollout: Jimeng AI (jimeng.jianying.com) and the Pro version of Doubao — both China-facing surfaces
- API status: BytePlus ModelArk published the model card `dreamina-seedance-2-5-260628` and a token rate card (~$6.40 / $10.70 per million tokens), but the console displays "The Model Playground and API access will be available soon" and the cURL sample still calls the 2.0 endpoint
- Native 4K, DCC-tool (Maya/Blender) integrations, and a 1080p rate row were NOT confirmed in the launch post
Source: ByteDance Seed — Seedance 2.5 official launch blog (Jul 31, 2026)
Seedance 3.0
Speculative · Not announcedNo announced release date
ByteDance has not announced a Seedance 3.0 release date. As of August 2, 2026, no public roadmap, no specifications, no timeline, and no Volcengine Ark placeholder for Seedance 3.0 exists. The newest official Seedance generation is 2.5.
- Based on ByteDance's historical cadence (1.0 → 2.0: 8 months; 2.0 → 2.5 launch: 5.5 months), a 3.0 release would not be expected before 2027 at the earliest
- Any "Seedance 3.0 release date" or specifications quoted today are speculation, not reporting
- Sites selling "Seedance 3.0 access" (e.g. seedance3ai.net and similar) are unaffiliated with ByteDance
2.5 API is still "coming soon" — build on 2.0 today
Seedance 2.5 launched on Jimeng AI and Doubao Pro on July 31, 2026, but the BytePlus ModelArk endpoint is not yet callable. 2.5 keeps the same unified multimodal audio-video architecture as 2.0, so prompts, camera direction, and reference-input patterns you learn on 2.0 will carry forward. Start a $15 Mini Pack now, get 300 credits and 12 months to use them.
12-month credit validity · no subscription · cancel anytime
Side-by-side capability comparison
Capabilities for Seedance 1.0, 2.0, and the officially launched 2.5 across seven dimensions. 2.5 entries reflect the ByteDance Seed launch post (July 31, 2026) and the BytePlus ModelArk model card and rate card published the same day.
| Capability | Seedance 1.0 | Seedance 2.0 (current API) | Seedance 2.5 (launched · API pending) |
|---|---|---|---|
| Max clip duration (single pass) | 5-8 seconds | 4-15 seconds | 30 seconds (confirmed by launch post, Jul 31, 2026) + multi-round extension for longer sequences |
| Max resolution | 720p | Up to 2K (native 480p/720p/1080p/4K rate rows documented) | 480p and 720p in the published rate card; 1080p and 4K not confirmed in launch materials |
| Native audio | No | Yes (stereo, BGM + SFX + voiceover + lip-sync) | Yes — same unified multimodal audio-video architecture as 2.0, extended to 30-second clips |
| Multimodal reference input | Text + single image | @-reference: 9 images + 3 videos + 3 audio | Up to 30 images + 10 videos + 10 audio clips per request; adds clay-render, motion, and creative reference types |
| Region-level / timestamp editing | No | No | Timestamp-level and camera-perspective editing (confirmed by launch post) |
| API access | None (consumer only) | Public on Volcengine Ark (since April 2, 2026); `dreamina-seedance-2-0-260128` on BytePlus ModelArk | Not callable as of Aug 2, 2026 — `dreamina-seedance-2-5-260628` listed but Model Playground and API "coming soon" |
| Pricing model | Free inside Doubao / Jimeng | Pay-per-token (~$0.14/sec generated at 720p) | Rate card published: ~$6.40 / $10.70 per million tokens (with / without video input); ~$0.103/s at 480p and ~$0.231/s at 720p per BytePlus ModelArk documentation |
What changed: Seedance 1.0 → 2.0
The 1.0 → 2.0 jump (8 months) was the biggest leap in the Seedance line so far. 1.0 maxed out at 5-8 second clips, 720p, with no native audio — clips were silent by default and had to be dubbed in post. 2.0 added native synchronized stereo audio in a single pass: background music, sound effects, voiceover, and lip-sync, all generated alongside the video.
The other 2.0 headline change was the @-reference multimodal input: a single generation request can now reference up to 9 images, 3 videos, and 3 audio clips simultaneously. This is what makes 2.0 viable for consistent-character storytelling and product-shot recreation — both of which 1.0 struggled with because it could only condition on a single image at a time.
The third big change was distribution: 1.0 was consumer-only (Doubao, Jimeng), with no public API. 2.0 opened a pay-per-token public-beta API on Volcengine Ark on April 2, 2026, at roughly $0.14 per second of generated video. That made 2.0 the first Seedance version a non-Chinese developer could integrate into a product without going through Chinese consumer accounts.
What shipped: Seedance 2.0 → 2.5
Seedance 2.5 was officially launched by ByteDance Seed on July 31, 2026 with the post "One-take Creation, Flexible Referencing: Introducing Seedance 2.5." Same-day TechNode coverage confirmed the scope: 30-second high-quality audio-video clips in a single pass with multi-round extension for longer sequences, reference input of up to 30 images + 10 videos + 10 audio clips per request, timestamp-level and camera-perspective editing, and improved shot transitions.
The rollout started on Jimeng AI and the Pro version of Doubao (both China-facing consumer surfaces). BytePlus ModelArk published the 2.5 model card dreamina-seedance-2-5-260628 and a token rate card at ~$6.40 / $10.70 per million tokens (with / without video input), but its own console displays "The Model Playground and API access will be available soon" above the request example, and its cURL sample still calls the 2.0 endpoint. As of August 2, 2026, no callable 2.5 API endpoint exists — third-party reports from July 16, 2026 claiming an already-open BytePlus 2.5 API are not supported by any official BytePlus page checked against this timeline.
The 2.5 release is a capability expansion built on the 2.0 architecture, not a redesign. The unified multimodal audio-video pipeline, prompt syntax, and reference-input mechanism all carry forward. That makes 2.0 prompt-engineering work and workflow investment effectively forward-compatible on day one of the eventual ModelArk endpoint opening.
What is still not confirmed for 2.5 in first-party materials as of August 2, 2026: a native 4K rate row (only 480p and 720p appear in the published ModelArk rate card, though 2.0 documents 4K), a callable ModelArk endpoint, exact per-token pricing under enterprise volume, and named DCC-tool (Maya/Blender) integrations. We will update this section the moment the endpoint opens.
See also: Seedance 2.5 detailed entity page
Try Seedance 2.0 right here
Seedance 2.1 release date
Seedance 2.1 was never released. ByteDance jumped from 2.0 directly to 2.5.
As of August 2, 2026, no Seedance 2.1 model exists. A May 2026 Pandaily report cited an unnamed source describing internal "2.1" improvements, but ByteDance never published a 2.1 announcement, a Volcengine Ark 2.1 endpoint, or a 2.1 release date. It was superseded by the Seedance 2.5 announcement on June 23, 2026 (Volcano Engine FORCE) and the official 2.5 launch on July 31, 2026. The newest official generation is Seedance 2.5; the current callable API model is still Seedance 2.0 (released February 12, 2026).
If you arrived here searching "Seedance 2.1 release date," the short answer is: the version label you are looking for is 2.5, not 2.1. The ~20% quality lift the Pandaily 2.1 rumor cited overlaps directionally with what ByteDance eventually shipped in 2.5 (30-second single-pass clips, expanded reference pool, timestamp editing), but the label 2.1 was never published as a first-party endpoint. Sites advertising "Seedance 2.1 release date" or "Seedance 2.1 access" are speculating unless they cite a first-party endpoint.
Last reviewed . We update this section immediately if ByteDance publishes any new 2.1-specific information.
Seedance 3.0 release date
Seedance 3.0 has no announced release date.
As of August 2, 2026, ByteDance has not published any roadmap, specifications, or timeline for Seedance 3.0. There is no Volcengine Ark placeholder, no BytePlus ModelArk model card, no ByteDance Seed research blog announcement, and no consumer surface for a Seedance 3.0 model. The newest official Seedance generation is Seedance 2.5 (officially launched July 31, 2026 on Jimeng AI and Doubao Pro; ModelArk API pending). The current callable API model is still Seedance 2.0 (released February 12, 2026).
Based on ByteDance's release cadence — 1.0 (June 2025) → 2.0 (February 2026, 8 months) → 2.5 launched (July 31, 2026, 5.5 months after 2.0) — a Seedance 3.0 release would not be expected before 2027 at the earliest. Any "Seedance 3.0 release date" you see quoted today is speculation, not reporting. Sites offering "Seedance 3.0 access" (such as seedance3ai.net and similar) are unaffiliated with ByteDance and are not running an official Seedance 3.0 model — because no such model exists publicly.
This page is reviewed every 14 days and updated immediately when ByteDance publishes any Seedance 3.0 information. Last reviewed .
Glossary
Reference terms used across the Seedance version history.
- Seedance
- ByteDance's line of video generation models. First released in June 2025 (1.0); the current production version is 2.0 (Feb 2026). Distributed inside the broader Doubao AI product family.
- ByteDance
- Chinese internet company that operates TikTok, Douyin, and the Doubao AI family. Seedance is developed by the ByteDance Seed research team.
- Doubao (豆包)
- ByteDance's flagship consumer AI assistant brand, covering chat, image, video, and voice. Seedance is the video generation model line inside Doubao.
- Volcengine Ark
- ByteDance's enterprise cloud platform (operated by Volcano Engine) where the Seedance 2.0 API is published. Pay-per-token billing; the canonical source for official Seedance 2.0 documentation.
- Jimeng (即梦)
- ByteDance / Jianying's creative tool for video and image generation. Uses Seedance internally. Chinese-language interface only.
- @-reference (multimodal input)
- The input system introduced in Seedance 2.0 that lets a single generation request reference up to 9 images, 3 videos, and 3 audio clips simultaneously, alongside the text prompt.
- Native synchronized audio
- Seedance 2.0's ability to generate video and matching stereo audio (BGM, sound effects, voiceover, lip-sync) in a single pass — no post-production audio sync required.
- Diffusion transformer
- The neural architecture family Seedance 2.0 belongs to: a transformer model that progressively denoises latent video tokens. Used by most leading 2026-era video models.
- Pro suffix (retired)
- Earlier Seedance versions used a "Pro" suffix (1.0 Pro, 1.5 Pro) to mark the higher-quality variant. ByteDance retired the Pro suffix with the 2.0 release — "Seedance 2.0" is the only 2.x naming.
- Knowledge cutoff (Seedance 2.0)
- Approximately Q4 2025 to Q1 2026. Subjects, brands, or events that emerged after early 2026 may be missing from the visual training data.
People also ask
The questions Google and Copilot users most often pair with "Seedance 2.5", "Seedance 3.0 release date", and "latest Seedance version" queries.
What is the latest version of Seedance?
Seedance 2.5, officially launched by ByteDance Seed on July 31, 2026 with the launch post "One-take Creation, Flexible Referencing: Introducing Seedance 2.5." It is live today on Jimeng AI and the Pro version of Doubao. BytePlus ModelArk published the model card `dreamina-seedance-2-5-260628` but marks API access as "coming soon" — so the current callable Seedance API model as of August 2, 2026 is still Seedance 2.0 (released February 12, 2026; Volcengine Ark public beta since April 2, 2026).
When was Seedance 2.5 released?
Seedance 2.5 was announced at the Volcano Engine FORCE 2026 conference on June 23, 2026 and officially launched by ByteDance Seed on July 31, 2026. Rollout started the same day on Jimeng AI and Doubao Pro, with BytePlus ModelArk API access marked "coming soon." As of August 2, 2026, no 2.5 API endpoint is callable yet.
Is Seedance 2.5 available via API?
Not yet. As of August 2, 2026, BytePlus ModelArk lists the 2.5 model card (`dreamina-seedance-2-5-260628`) with a published token rate card, but the Model Playground and API access are marked "coming soon." The ModelArk cURL sample still calls the 2.0 endpoint. Any third-party page claiming a callable Seedance 2.5 API should be checked against the ModelArk model list before integration.
What's the difference between Seedance 2.0 and 2.5?
Seedance 2.5 doubles the single-clip ceiling from 15 to 30 seconds, expands reference input from 9 images + 3 videos + 3 audio (2.0) to up to 30 images + 10 videos + 10 audio clips per request, adds clay-render / motion / creative reference types, and introduces timestamp-level editing. The BytePlus rate card lists 480p and 720p only (not 1080p or 4K), priced roughly 50% above the equivalent 2.0 tier. Seedance 2.0 remains the only callable Seedance API model as of August 2, 2026.
What is the Seedance 3.0 release date?
Seedance 3.0 has no announced release date. As of August 2, 2026, ByteDance has not published any roadmap, specifications, or timeline for Seedance 3.0. Based on ByteDance's historical cadence — 1.0 (June 2025) → 2.0 (February 12, 2026, 8 months) → 2.5 (launched July 31, 2026, 5.5 months after 2.0) — a 3.0 release would not be expected before 2027 at the earliest. Any "Seedance 3.0 release date" you see quoted today is speculation, not reporting.
When will Seedance 3.0 be released?
No release date has been announced. ByteDance has not confirmed Seedance 3.0 is in development, has not published a roadmap, and has no Volcengine Ark placeholder for a 3.0 model as of August 2, 2026. The newest official ByteDance video generation is Seedance 2.5 (launched July 31, 2026). Any Seedance 3.0 release date forecast you encounter is inference, not reporting.
Is Seedance 3.0 out yet?
No. Seedance 3.0 has not been released and has not been announced. The newest official Seedance generation is 2.5, launched July 31, 2026 on Jimeng AI and Doubao Pro. Sites offering "Seedance 3.0 access" (such as seedance3ai.net and similar) are unaffiliated with ByteDance and are not running an official Seedance 3.0 model — because no such model exists publicly.
What happened to Seedance 2.1?
Seedance 2.1 was reported by Pandaily in May 2026 as in development with a target ~20% quality lift over 2.0, but it was never formally launched. It was superseded by the Seedance 2.5 announcement on June 23, 2026 and the official 2.5 launch on July 31, 2026. No Volcengine Ark page, no API, and no consumer surface for "Seedance 2.1" ever materialized.
Where can I use Seedance 2.5 today?
Two ByteDance-owned consumer surfaces confirmed by the official launch post: Jimeng AI (jimeng.jianying.com) and the Pro version of Doubao — both China-facing consumer products. There is no official English-language ByteDance consumer surface for 2.5, and no callable BytePlus ModelArk endpoint. Third-party operators (including seedance2-video.com) currently run the Seedance 2.0 API on Volcengine Ark and will update to 2.5 only once ModelArk opens the endpoint.
Sourced claims
Every load-bearing fact on this page, with its public source. Verified as of 2026-08-02.
Seedance 2.5 was officially launched by ByteDance Seed on July 31, 2026 with the post "One-take Creation, Flexible Referencing: Introducing Seedance 2.5." Confirmed capabilities: 30-second single-pass audio-video generation with multi-round extension; reference input of up to 30 images, 10 videos, and 10 audio clips per request; clay-render / motion / creative reference types; timestamp-level editing. Rollout: Jimeng AI and the Pro version of Doubao; BytePlus ModelArk API access "coming soon."
Source: ByteDance Seed · 2026-07-31
TechNode's same-day report (July 31, 2026) confirms the Seedance 2.5 launch: 30-second single-pass generation, multi-turn extension, up to 30 images + 10 videos + 10 audio references per request, rollout on Jimeng AI and Doubao Pro, and API access "expected to arrive soon on Volcano Engine's Ark platform."
Source: TechNode · 2026-07-31
Seedance 2.0 was officially launched by ByteDance on February 12, 2026, with native synchronized audio, 4-15 second clip generation, and an @-reference multimodal input system supporting up to 9 images, 3 videos, and 3 audio clips per request.
Source: ByteDance Seed · 2026-02-12
The Seedance 2.0 API became publicly accessible on Volcengine Ark starting April 2, 2026, with pay-per-token pricing of approximately 46 CNY (~$6.40) per million tokens for pure generation, equating to roughly $0.14 per second of output video. The 2.0 endpoint (`dreamina-seedance-2-0-260128`) remains the current callable Seedance API model as of August 2, 2026.
Source: TechNode · 2026-03-05
Pandaily reported Seedance 2.1 development in May 2026, but the label was not formally launched. It was superseded by the Seedance 2.5 announcement on June 23, 2026 (Volcano Engine FORCE) and the official 2.5 launch on July 31, 2026.
Source: Pandaily · 2026-05-15
ByteDance has not published any specifications, release date, or roadmap for Seedance 3.0 as of August 2, 2026. The current ByteDance Seed model catalog highlights Seedance 2.5 as the newest generation; no Seedance 3.0 placeholder has been identified on the ByteDance Seed research site or on Volcengine Ark / BytePlus ModelArk.
Source: ByteDance Seed (official research site) · 2026-08-02
The Seedance product line, including all prior versions (1.0, 1.5 Pro), the current callable API model (2.0), and the newest generation (2.5), is developed by the ByteDance Seed research team and distributed inside the Doubao AI product family.
Source: Wikipedia · 2026-04-10
Frequently asked questions
How often does ByteDance release a new Seedance version?
Based on the public release history, roughly every 4-8 months and accelerating. Seedance 1.0 launched in June 2025; an interim 1.5 Pro followed in late 2025; Seedance 2.0 launched on February 12, 2026 (8 months after 1.0); Seedance 2.5 was announced on June 23, 2026 and officially launched on July 31, 2026 (5.5 months after 2.0). ByteDance has not committed to a public release schedule.
Should I wait for the Seedance 2.5 API or use 2.0 today?
Use 2.0 today if you need any callable Seedance API. As of August 2, 2026, the BytePlus ModelArk 2.5 endpoint is still marked "coming soon" — there is no way to build a production integration on 2.5 yet. The 2.0 endpoint (`dreamina-seedance-2-0-260128`) is documented, stable, and priced. Prompt-engineering work on 2.0 also carries forward: subject, camera, audio, and reference-input patterns transfer, though the exact request schema for 2.5 will only be known when the ModelArk endpoint opens.
Will my Seedance 2.0 prompts and workflows still work on 2.5?
2.5 keeps the same unified multimodal audio-video architecture as 2.0, so single-shot text-to-video, image-to-video, and audio-referenced prompts should carry forward directionally. The concrete changes to plan for: 30-second single-pass clips (vs 15s), a much larger reference pool (30 images + 10 videos + 10 audio vs 9 + 3 + 3), and new reference types (clay-render, motion, creative). Timestamp-level editing is a new capability, not a replacement for existing prompt patterns. Keep prompts modular so subject / camera / audio blocks can be recombined once the 2.5 API opens.
How is Seedance 2.5 priced compared to 2.0?
BytePlus ModelArk published a 2.5 token rate card at roughly $6.40 per million tokens (with video input) and $10.70 per million tokens (without video input). Third-party integrators tracking the rate card have derived approximate per-clip prices of $0.514 at 480p ($0.103/s) and $1.156 at 720p ($0.231/s) — roughly 50% above the equivalent 2.0 tier. No 1080p or 4K rate row has been published for 2.5. These prices reflect the ModelArk rate card as of August 2, 2026; the endpoint itself is not yet callable, and Volcengine (¥42 / ¥70 per million tokens for `doubao-seedance-2-5-260628`) uses CNY billing.
Where does Seedance 2.5 launch the public API?
BytePlus ModelArk (docs.byteplus.com/docs/ModelArk) — the international BytePlus endpoint — is the channel ByteDance Seed points to in the July 31, 2026 launch post. The Volcengine Ark console (volcengine.com/product/ark) hosts the Chinese `doubao-seedance-2-5-260628` model card. As of August 2, 2026, both consoles publish model cards and pricing but keep API access marked "coming soon."
Is "Seedance Pro 2.0" or "Seedance 2.7" a real version?
No. The official ByteDance Seedance line as of August 2, 2026 is: 1.0, 1.5 Pro, 2.0 (still the current callable API), and 2.5 (newest generation, launched July 31, 2026). There is no "Seedance 2.0 Pro," no "Seedance 2.7," and no "Seedance 3.0." The "Pro" suffix was retired with the 2.0 release. Any site selling a variant with a different number or suffix is using marketing names that do not match the official ByteDance product line.
If Seedance 3.0 has not been announced, why does this page mention it?
Because users actively search for "Seedance 3.0 release date" — and an honest answer ("no announced release date, no public information; newest generation is 2.5") is more useful than letting that query route to misinformation. This page is the canonical "no, there is no Seedance 3.0 release date" answer for the 3.0 query, with the same source-cited treatment applied to confirmed versions like 2.0 and 2.5.
How is this page kept up to date?
This page is reviewed at least every 14 days and updated immediately when ByteDance publishes new Seedance information. The "Last updated" date in the page header and the schema.org dateModified field both reflect the most recent factual update. The most recent update on August 2, 2026 records the Seedance 2.5 official launch (July 31, 2026), the confirmed 30-second / 30-image capability envelope, the BytePlus ModelArk "coming soon" API status, and the published 2.5 rate card.
Stay current on Seedance
This page is reviewed every 14 days and updated immediately when ByteDance ships anything new. Bookmark it, or jump into the 2.0 generator and get a head start.
$15 to start · 12-month credits · no subscription required