modelstringSelects the route — one of the currently available model IDs above.
Access ByteDance Seedance 2.5 through EvoLink's unified video API. Test text-, image-, and reference-to-video, video edit, and video extend first.
Test Seedance 2.5 output quality, estimate credit usage, and generate a sample before you formally integrate.
Create a Sample
Pick a video mode, enter your prompt, and set the parameters.Off here. The API still supports it — send content_filter: false to relax content restrictions (billed at +10%). Illegal and prohibited content is always blocked.
Results are kept for 24 hours — download anything you want to keep.
Start from the cost of a single sample and pick a testing budget. Expand for detailed rates.
1080p -28%Limited-time offer — Seedance 2.5 1080p is 28% off through September 17, 2026 (UTC+8). Discount included below; 480p and 720p are unchanged.
First Test Cost
Text-to-Video, 4s at 480p — the cheapest possible request, no reference filesOne 4s 480p sample
About 18 tests with $10 in credits.
Use 480p for lower-cost prompt iteration, 720p for quality checks, and 1080p for final output.
Testing Budget Guide
At the first-test settings: 4s at 480pGood for a first validation.
Good for prompt iteration.
Good before production integration.
Estimated cost per run for the listed parameters. Final cost may vary by mode, references, and add-ons.
| Mode | 480p | 720p | 1080p-28% | Billing |
|---|---|---|---|---|
| Text-to-VideoStandard | $0.138/s9.3419 cr/s | $0.296/s20.09 cr/s | $0.532/s36.162 cr/s | Output video seconds |
| Image-to-VideoStandard | $0.138/s9.3419 cr/s | $0.296/s20.09 cr/s | $0.532/s36.162 cr/s | Output video seconds |
| Reference-to-VideoImage or audio reference | $0.138/s9.3419 cr/s | $0.296/s20.09 cr/s | $0.532/s36.162 cr/s | Output video seconds |
| Reference-to-VideoVideo reference | $0.084/s5.7056 cr/s | $0.180/s12.2349 cr/s | $0.324/s21.9865 cr/s | Input + output seconds |
| Video EditVideo reference | $0.084/s5.7056 cr/s | $0.180/s12.2349 cr/s | $0.324/s21.9865 cr/s | Input + output seconds |
| Video ExtendVideo reference | $0.084/s5.7056 cr/s | $0.180/s12.2349 cr/s | $0.324/s21.9865 cr/s | Input + output seconds |
Billed as output duration (seconds) × the per-second resolution rate; audio adds no extra cost. When a reference video is provided (Reference-to-Video with video input, Video Edit, Video Extend), the lower video-reference rate applies and input duration is billed too: billable input = max(total input video duration, output duration), cost = (billable input + output) × rate. Auto duration (duration: -1) and Video Edit settle on the actual output length, rounded up to the next 0.1s.
Auto duration: the model picks the length, so credits are reserved at the 30s cap — about 602.7 credits (~$8.864) at 720p. You are only charged for the seconds actually generated; the rest is refunded when the task finishes.
| Item | Use Case | Price | Credits |
|---|---|---|---|
| Web SearchPer request | Use when the prompt needs current events, products, or real-world references. | $0.0006 / request | 0.04 cr |
Most first tests only pay the video generation fee. Logged-in users can view live pricing in the console.
Seedance 2.5 is a multimodal video generation model from ByteDance / BytePlus. Its API opened on August 7, 2026. The model supports text-to-video, image-to-video, reference-to-video, video-edit, and video-extend routes for generating 4–30 second videos at 480p, 720p, or 1080p. Use the routes currently listed on this page for EvoLink requests. It is built for ads, UGC, dialogue, and story workflows that need longer shots, multimodal references, character or product consistency, and synchronized audio.
More reference files do not automatically produce a better result. First-pass usability depends on choosing the right route and giving character, product, scene, motion, and audio references clear roles.
seedance-2.5-text-to-videoGenerate 4–30 second videos from a script, ad brief, or storyboard, with synchronized audio and optional web search.
seedance-2.5-image-to-videoUse one image as the first frame or two images as first-and-last-frame constraints for product, character, and composition-led shots.
seedance-2.5-reference-to-videoCombine image, video, and audio references when character, product, camera motion, action, and sound must be controlled together.
seedance-2.5-video-editEdit an uploaded video with a prompt — remove, replace, or restyle elements. Output length and ratio follow the input video.
seedance-2.5-video-extendExtend an uploaded video forward or backward with continuous motion, up to 30 seconds of new footage.
Keep the entrance, main action, and ending in one ad, dialogue, or story clip.
Use up to 30 images, 10 videos, and 10 audio tracks.
Evaluate visuals, dialogue, sound effects, and pacing in the same task.
Manage long-running jobs with a task ID, polling, and HTTPS callbacks.
modelstringSelects the route — one of the currently available model IDs above.
promptstringThe generation or edit instruction. Required on every route.
durationinteger4 through 30, or -1 for automatic length. Video Edit always follows the input video.
qualitystring480p, 720p, or 1080p. Billing follows the quality tier and output seconds.
aspect_ratiostringadaptive, 16:9, 9:16, 1:1, 4:3, 3:4, or 21:9. Image-to-Video and Edit/Extend accept adaptive only.
image_urlsstring[]Image-to-Video takes 1–2 first/last-frame images; Reference-to-Video takes up to 30 reference images.
video_urlsstring[]Up to 10 reference videos; Edit and Extend require the input video here.
audio_urlsstring[]Up to 10 audio tracks for voice, music, or sound reference.
generate_audiobooleanSwitch for synchronized audio in the output video.
callback_urlstringPublic HTTPS URL that receives the final task state; keep the task ID as a polling fallback.
The available routes post to the same endpoint and share this parameter set; each route enforces its own whitelist and rejects unsupported fields at validation. A request can carry at most 50 media references in total. Full constraints and response fields are in the API Reference.
Use the EvoLink API for product integration and batch jobs, or call it from Codex, Claude, or Gemini for fast creative and development workflows. Both paths share the same EvoLink API key, balance, model routes, and task history.
Best for: product backends, batch jobs, automated workflows
Call EvoLink’s unified video API from your server and control the model ID, parameters, task queue, callbacks, and result storage.
Best for: creative and development tasks in Codex, Claude, and Gemini
Give the Agent your output goal, assets, and acceptance criteria. It can choose the route, assemble the request, track the task, and return the result without requiring you to hand-code every step.
Illustrative clips for building a prompt and evaluation set. These are not presented as verified Seedance 2.5 API outputs.
This example shows the shortest runnable flow: create a task, save the returned task ID, then poll or wait for an HTTPS callback. Open the API tab for the complete parameter and response reference.
curl -X POST https://api.evolink.ai/v1/videos/generations \
-H "Authorization: Bearer $EVOLINK_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "seedance-2.5-text-to-video",
"prompt": "A cinematic product film with slow camera movement",
"duration": 5,
"quality": "720p",
"aspect_ratio": "adaptive",
"generate_audio": true,
"content_filter": true
}'
# Save the returned task ID, then query:
curl https://api.evolink.ai/v1/tasks/{task_id} \
-H "Authorization: Bearer $EVOLINK_API_KEY"Check that duration is 4–30, quality is 480p, 720p, or 1080p, and the asset count and file types match the selected route.
Check the Authorization bearer token and confirm the available balance and task billing in the console.
Review likeness rights, brand assets, readable text, sensitive content, and input-asset compliance.
A Seedance 2.5 task that reaches a final failed state is not charged. Keep the task ID, inspect the final state, and identify whether the cause is validation, review, or execution before retrying.
callback_url must be a public HTTPS URL, not localhost or a private network address. Always keep the task ID as a polling fallback.
Seedance 2.5 is not the default upgrade for every task. Use the same brief, input assets, and acceptance criteria to compare usable-shot rate, reruns, and total project cost.
| Decision factor | Seedance 2.5 | Seedance 2.0 | MiniMax H3 |
|---|---|---|---|
| Best suited to | 16–30s shots, multimodal references, UGC, ads, and dialogue | Mature Seedance workflows, shorter shots, and fast iteration | 2K short clips, action, and high-detail visuals |
| Duration | 4–30 seconds | 4–15 seconds | 4–15 seconds |
| Resolution | 480p / 720p / 1080p | 480p / 720p / 1080p (higher on some Standard tiers) | 2K / 768p |
| Inputs | Text, first/last frames, image/video/audio references | Text, images, and multimodal references | Text, first/last frames, and multimodal references |
| Cost decision | Measure cost per usable shot and whether longer clips reduce editing and reruns | Keep the current baseline and use the same brief to test whether upgrading is worthwhile | Compare usable 2K output and live pricing with the same brief |
Three facts that shape the integration: routing, task lifecycle, and billing.
Every route posts to /v1/videos/generations. The model field selects the route, and each route enforces its own parameter whitelist — unsupported fields are rejected at validation instead of being silently ignored.
A request returns a task ID immediately. Poll every 5–10 seconds or register a public HTTPS callback_url, and keep the task ID either way. A task whose final state is failed is not charged.
Cost = output duration × the per-second rate of the selected quality tier. Reference-to-Video, Video Edit, and Video Extend also bill the input video duration — see the Pricing section for rates.
The hard part of a production Seedance 2.5 API integration is not one request. It is managing model choice, cost, references, asynchronous jobs, callbacks, failures, and fallback routes after launch. EvoLink brings these into one video API and console so teams can test first, integrate second, and route by usable-shot rate and total project cost.
Available Seedance 2.5 routes share EvoLink authentication, balance, and task infrastructure. Your server stores one EVOLINK_API_KEY and selects the appropriate model ID for each input and task.
When you also need Seedance 2.0, MiniMax H3, or another video model, you do not have to maintain separate provider accounts, keys, balances, and authentication code. Keep the same task and billing framework and change routes by duration, quality, and use case.
Use a real ad brief, character image, product image, camera reference, and audio asset in Playground before writing your production integration. That is more reliable than judging the model with one demo prompt.
Track usable-shot rate, reruns, and manual rework—not only price per second. Surface character drift, logo deformation, motion blur, reference conflicts, and audio-video sync issues before you lock the route, prompt structure, and acceptance criteria.
EvoLink does not charge a Seedance 2.5 task whose final state is failed. Save the task ID and verify the final status, failure reason, and billing record in the console instead of treating “request created” as “video generated.”
Video generation is asynchronous. Production systems should use polling or a public HTTPS callback, handle missing or duplicate callbacks and timeouts, and retry only after identifying a validation, review, or execution failure.
Product backends can run batch tasks through the EvoLink API, while creative and development teams can ask Codex, Claude, or Gemini to assemble requests, track status, and return results. Both methods share the API key, routes, balance, and task history.
Developers, operators, and content teams can collaborate on one generation foundation instead of maintaining separate accounts and cost records for code and Agent workflows. Copy the Agent prompt on this page, complete the generation goal, and paste it into your Agent.
Compare Seedance routes under one EvoLink account and API key, then choose by shot duration, reference workflow, speed, and cost.

Seedance 2.0
A mature Seedance production baseline for shorter shots, multimodal references, and existing 2.0 workflows.
View API
Seedance 2.0 Mini
A lightweight, lower-cost Seedance route for drafts, prompt iteration, and frequent batch jobs.
View API
Seedance 1.5 Pro
A previous-generation cinematic route for migrating older projects and comparing motion quality.
View APIWhen Seedance 2.5 is not the default, use the same brief to compare resolution, shot duration, audio, motion stability, and total project cost.

MiniMax H3
For detailed 2K clips of 4–15 seconds; compare its usable-shot rate with Seedance 2.5 using the same brief.
View API
Kling 3.0
For creative generation, action, and multi-shot video; a production fallback outside the Seedance family.
View API

Topaz Video Upscale
When a Seedance shot already has usable motion and composition but looks soft or compressed, use Topaz for detail recovery and 1x, 2x, or 4x upscaling.
View APIHigh-intent questions about availability, model IDs, pricing, limits, model selection, and troubleshooting.
Yes. Choose from the available routes currently shown on this page. Test the output in Playground, then integrate with the same EvoLink API key.
The mode cards and code example above show the currently available model IDs. Choose the route by input and task type; do not send media URLs to the text-to-video route.
Yes. After setting EVOLINK_API_KEY in the local environment, Codex, Claude, or Gemini can choose a model ID, submit the task, poll its status, and return the video. Never place the API key in a prompt or frontend code.
Billing depends on video duration and resolution; video references may also include input duration. Use the live estimate in Pricing and Playground instead of relying on a hard-coded price in the Readme.
No. EvoLink does not charge a task whose final state is failed. Keep the task ID and verify the final status, failure reason, and billing record in the console.
Seedance 2.5 supports 4–30 second output at 480p, 720p, or 1080p. Validate composition, motion, and consistency at 480p, check quality at 720p, then move an approved setup to 1080p.
Reference-to-video accepts up to 30 images, 10 videos, and 10 audio tracks, with no more than 50 assets per request. Give each character, product, scene, motion, and audio reference a clear role.
Yes. You can enable generated audio or supply an audio reference in reference-to-video mode. Keep a fixed evaluation set for dialogue, lip sync, and audio-video pacing.