Seedance 2.5 · AnyCap integration · Catalog and editing schema checked October 26, 2026 · Output trial recorded September 22, 2026
Seedance 2.5 API access through AnyCap
Use seedance-2.5 to generate video from text, images, first and last frames, or mixed media references through AnyCap. Choose the mode before preparing your assets: reference fields and input limits differ. The four generation modes support 480p, 720p, and 1080p. Use edit-video to revise one existing clip while preserving its duration and aspect ratio. Use AnyCap's HTTP task API or CLI to submit a draft, resume the same task, and download its video.
One five-second text-to-video trial
We ran the desk-lamp command shown below through AnyCap CLI 0.6.2 against the production API: text-to-video, 5 seconds, 720p, 16:9, generate_audio=false. The delivered file is 1280 × 720 at 24 fps, approximately 5.04 seconds, with no audio stream. The CLI reported 59 credits for this task; this is one observed charge, not a fixed price quote.
- The output shows an olive cylindrical shade on a metal stand, with a camera push-in and no visible lettering in the inspected frames.
- The framing becomes much tighter: the shade is already cropped at the top around two seconds, and the base leaves the frame later. If the full product must stay visible, make that requirement explicit and review a new draft.
- This was one prompt and one output, reviewed with sampled frames and video analysis. It does not establish model-wide quality, repeatability, or a generation-time benchmark.
From request to a saved video
Start with one short draft, keep its task ID, and review the delivered clip before generating more variants.
Choose the inputs
Start from a text prompt, reference images, or first and last frames. Choose the mode, duration, resolution and sound before submitting.
Submit one draft
Confirm account access and current pricing first. Keep the returned task ID so an interrupted session can return to this generation.
Wait for delivery
An accepted request is still a pending job. Wait until the finished video is available. If waiting is interrupted, resume the same task.
Download and review
Save the finished clip and actual charge. Check framing, continuity and reference details. If the local save fails, download that result again.
Check current pricing and your balance before submitting. For API access, create a key in API keys. If submission loses its response, check the recorded request before paying for another generation.
For installation, supported controls and file delivery, read the video documentation.
Edit existing footage without starting again
Choose edit-video for an existing clip and a text instruction. Supply exactly one source video through videos. Optional image references guide the edit; at most one audio reference is supported on this route. Keep videos, images and audio in separate fields.
Editing follows the source duration and aspect ratio. Omit those controls or use duration=-1 and aspect_ratio=adaptive; conflicting explicit values are rejected. Draft submission is not supported on this editing route. Check the live schema for current limits before uploading.
The recorded desk-lamp trial demonstrates text-to-video only. It is not an editing test or evidence that every account has access to this route.
Start with one Seedance 2.5 draft
A five-second draft saved as seedance-desk-lamp.mp4. Compare with the recorded trial; preserve the task ID if generation is pending.
Install the CLI and sign in once. If you are already configured, continue with the access and cost checks.
curl -fsSL https://anycap.ai/install.sh | sh
anycap login
anycap --version
anycap statusRun anycap status, inspect the current model schema, and check pricing and your balance. Generation is billed; model access and rates can change.
anycap video models seedance-2.5 schema --operation generate --mode text-to-videoRun this command after checking the schema and cost. Inspect the delivered file and actual charge.
anycap video generate --model seedance-2.5 --mode text-to-video --prompt "An olive desk lamp on a pale studio table. Slow camera push-in, steady lighting, no text." --param duration=5 --param resolution=720p --param aspect_ratio=16:9 --param generate_audio=false -o seedance-desk-lamp.mp4With CLI 0.6.5, expired video links are refreshed once by querying the same task, and transient downloads are retried. A pending task ID means the job is still running. Keep that ID and resume waiting after a local timeout or download failure; a new generation command creates a separate billed task.
anycap video tasks get <task-id>
anycap video tasks wait <task-id> -o seedance-desk-lamp.mp4Read the related workflow. For setup help, see installation and login.
Use the CLI for the recorded draft
The example requests five seconds at 720p without generated audio, a small first trial with explicit settings. These commands match the checked request fields.
Confirm availability and inspect the intended mode
anycap video models seedance-2.5
anycap video models seedance-2.5 schema --operation generate --mode text-to-videoExample: one short text-to-video draft
anycap video generate --model seedance-2.5 --mode text-to-video --prompt "An olive desk lamp on a pale studio table. Slow camera push-in, steady lighting, no text." --param duration=5 --param resolution=720p --param aspect_ratio=16:9 --param generate_audio=false -o seedance-desk-lamp.mp4For a short-lived agent session, add --no-wait, save the returned task ID, and resume waiting and downloading with anycap video tasks wait <task-id> -o seedance-desk-lamp.mp4. Use anycap video tasks get <task-id> for a single status check. Check the existing task before submitting another generation. A pending task is not a reason to create a duplicate. The default blocking command waits and downloads the result to the -o path.
Choose the mode from the inputs you have
Generation modes have a requested duration of 5–30 seconds, as checked against live schemas on October 5. Editing reflects the editing schema checked on October 26; inspect the live schema before use. For editing, and when multi-modal-reference includes a video, use duration=-1 to match its complete duration and aspect_ratio=adaptive to follow its aspect ratio; AnyCap uses those values when the fields are omitted.
| Mode | Reference fields | Resolution | Decision |
|---|---|---|---|
| edit-video | Exactly one source video; up to 30 image references and one audio reference | 480p, 720p, 1080p | Revise footage while preserving source duration and aspect ratio. |
| text-to-video | A prompt | 480p, 720p, 1080p | Start without reference media. |
| image-to-video | Up to 9 images; schema also exposes up to 3 video references | 480p, 720p, 1080p | Use image and video fields from the selected mode schema. |
| first-last-frame-to-video | Named first_frame and last_frame inputs | 480p, 720p, 1080p | Describe the transition between the two frames. |
| multi-modal-reference | Up to 9 images, 3 videos, and 3 audio references | 480p, 720p, 1080p | Combine the supported reference types in one request. |
Schema support confirms the request shape; it does not establish output quality or guarantee that every input combination succeeds.
Audio references and generated audio are separate controls. Supplying a reference clip guides the request; generate_audio controls whether output audio is requested. Output duration does not trim your reference audio.
For HTTP requests, reference fields need reachable media URLs or previously uploaded AnyCap references. A local path such as ./input.png is not a file upload in a JSON body. The CLI uploads supported local files when resolving --param values before submitting the request.
Estimate cost before expanding a draft into a batch
The AnyCap model listing marks Seedance 2.5 pricing as variable. Check current pricing for the selected settings. Model discovery and schema responses do not include a billable quote, so a numeric Seedance 2.5 cost cannot be inferred from them.
For a per-second rate, a five-second draft uses five times that rate before any applicable billing adjustments. Three separately billed five-second drafts represent fifteen seconds of generated output, even if you keep only one. Compare the actual charge from your first successful task with the estimate before scaling.
AnyCap modes and Dreamina features have different boundaries
Dreamina is a separate product surface. Its long-video, extension, or editing controls do not automatically become modes in the AnyCap API. The AnyCap catalog checked on October 26 lists four generation modes plus edit-video, but no separate Seedance 2.5 extension mode. For Dreamina-specific controls, consult the provider product page.
If your task needs structured shot-by-shot prompts, compare the Kling 3.0 multi-shot mode. For another Seedance version, inspect Seedance 2.0 on its own terms; do not assume the same parameters or prices.
Review the result against the brief
Check the full clip for subject continuity, camera movement, readable text, and any changes to reference details. A successful task response proves delivery, not suitability for publication. Keep the prompt, input files, output settings, and actual charge alongside the chosen draft.
Seedance 2.5 questions
Is Seedance 2.5 available through AnyCap?
The public AnyCap catalog checked on October 26, 2026 lists seedance-2.5 as active with four generation modes plus edit-video. Catalog availability is separate from a successful paid generation; your account also needs valid authentication and access. Inspect the live catalog before use.
How long can I generate through this integration?
The checked production schemas expose integer durations from 5 through 30 seconds. Multi-modal-reference also exposes -1 to match a complete reference video. Dreamina product features such as long-video generation or extension should not be assumed to be separate AnyCap modes.
Can I request 1080p in every mode?
The four generation schemas checked on October 5 list 480p, 720p, and 1080p. Editing has separate controls; do not assume generation options apply. Inspect the selected mode before submitting a request.
Can I use 30 reference images in image-to-video?
No. The checked production image-to-video and multi-modal-reference schemas both list a maximum of nine images. Multi-modal-reference also exposes up to three video and three audio references.
How much does Seedance 2.5 cost?
AnyCap lists variable pricing for this model. Confirm the rate for the intended mode and output settings before running a batch, then use the actual charge from a successful task to estimate revisions. A schema response does not provide a price quote.
Does this page provide the provider's native API?
AnyCap provides API and CLI access to its video generation service. This guide covers the available modes, a recorded output and the delivery workflow. Use AnyCap documentation for integration setup; ByteDance and Dreamina have separate product surfaces.
Verification and next steps
The four generation schemas were checked on October 5, 2026. The public catalog and edit-video schema were rechecked on October 26, 2026. The text-to-video trial shown above was recorded through AnyCap CLI 0.6.2 on September 22, 2026. Other modes and reference limits were checked from schemas only; no editing trial is claimed. Recheck the catalog before relying on these settings.
CLI version and updating · 0.6.5
Current release checked October 21, 2026: 0.6.5. Use the update command for your installation method, then verify the version and login in the same environment that runs your agent.
Direct binary installation
anycap updatenpm installation
npm install -g @anycap/cli@latestanycap --version
anycap statusVersion 0.6.5 retries transient downloads, recovers an expired video link by querying the same task, and saves multiple image outputs to numbered files. After a video download failure, resume the existing task instead of generating again.
anycap video tasks wait <task-id> -o result.mp4Dated examples retain the CLI version, output and charge from the original run. Updating the CLI does not revalidate an old result. Check the live model schema and current pricing before a new billed task.