Seedance 2.5 R2V
ByteDance's Seedance 2.5 R2V generates up to 30-second cinematic clips from images with precise reference control, multimodal input fusion, and native audio.
Overview
What is Seedance 2.5 R2V
Seedance 2.5 R2V is ByteDance's next-generation image-to-video model, released in July 2026. It generates up to 30-second audio-video clips in one pass using reference images, enabling cinematic continuity, strong character consistency, and precise creative control without stitching or drift.
Running it privately on Venice
On Venice, Seedance 2.5 R2V runs under an anonymized privacy tier — your reference images and prompts are not stored, profiled, or used for training. This means you retain full sovereignty over sensitive creative assets while accessing high-resolution, audio-enabled video generation in a permissionless environment. Pay only per clip, with no subscription lock-in.
Assessment
Strengths and limitations
- Native 30-second video generation in one pass: no stitching, no consistency drift between scenes.
- Strong reference-to-video (R2V) control: accurately preserves character, style, and motion from input images.
- Supports multimodal references: combine up to 30 images, 10 videos, and 10 audio clips in one generation.
- Generates native audio tracks synchronized with visuals, enhancing cinematic realism.
- High-resolution output (up to 1080p) with cinematic and photorealistic quality.
- Closed and proprietary: no open weights, so self-hosting or fine-tuning is not possible.
- Higher cost per second than Seedance 2.0, making it less ideal for rapid prototyping or A/B testing.
- No support for end-to-end encryption or TEE on Venice, limiting extreme-security use cases.
- Limited to 30-second maximum duration: not suitable for long-form content without external editing.
Capabilities
What it supports
- Image to video
- Native audio generation
Specifications
Datasheet
- Maker
- ByteDance
- Released
- July 2026
- Architecture
- Multimodal audio-video joint generation
- Resolutions
- 1080p, 720p, 480p
- Clip lengths
- 4s – 30s
- Mode
- image-to-video
- Aspect ratios
- 21:9, 16:9, 4:3, 1:1, 3:4, 9:16
- Audio
- Yes
- Prompt limit
- 15,000 chars
- Privacy on Venice
- Anonymized — prompts not stored
- Available on Venice since
- Aug 2026
- License
- Proprietary
API
Call it from your code
Venice exposes this model through the REST API. Queue a generation with the model id.
curl https://api.venice.ai/api/v1/video/queue \
-H "Authorization: Bearer $VENICE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "seedance-2-5-reference-to-video-basic",
"prompt": "Aerial drone shot over a misty mountain valley at golden hour"
}'
# Use the returned queue_id with https://api.venice.ai/api/v1/video/retrieve.
# Call /video/complete after downloading if needed.Pricing
What it costs on Venice
Pay per clip on Venice — price scales with resolution and duration (4s–30s), from $0.51.
New Venice accounts include a free daily allowance and 500 welcome credits — no credit card required.
Alternatives
How it compares
| Model | Best for | Max duration | Reference inputs | Open/closed | Price (Venice) |
|---|---|---|---|---|---|
| Seedance 2.5 R2V | Strongest in reference fidelity and cinematic continuity. | 30s | 30 images, 10 videos, 10 audio | Closed | from $0.51 |
| Wan 2.7 Enhanced | Open-weight alternative with strong image-to-video but less polished audio and motion. | — | — | Yes | from $0.68 |
| Kling O3 Pro | High-quality video but shorter native duration and weaker reference control. | — | — | No | from $0.46 |
| Vidu Q3 | Balanced performance but lacks Seedance 2.5's multimodal reference depth. | — | — | No | from $0.27 |
Strongest in reference fidelity and cinematic continuity.
Use cases
What it is good for
- 01Creating high-fidelity commercials or social ads with consistent characters and branding.
- 02Generating cinematic scenes from concept art or storyboards using image references.
- 03Producing music videos or short films with synchronized audio and visual continuity.
- 04Iterative creative workflows where reference control ensures brand or character fidelity.
- 05Marketing teams needing polished, audio-visual assets without production crews.
Prompting
Getting better results
Use high-quality, well-lit reference images for best character and style fidelity.
Limit reference inputs to the most essential — too many can dilute intent.
Specify aspect ratio and duration clearly in the prompt to match platform needs.
Use audio references to guide soundtrack tone, even if final audio will be replaced.
Start with 480p for testing, then scale to 1080p for final output to manage cost.
Version history
Predecessor with shorter clip length and weaker reference control.
Current — enhanced R2V, 30s native, multimodal input.
FAQ
Frequently asked questions
Seedance 2.5 R2V is ByteDance's advanced image-to-video model that generates up to 30-second cinematic clips using reference images, videos, and audio. It enables precise creative control, strong character consistency, and native audio generation without stitching.
On Venice, pricing starts at $0.51 for a 480p, 4-second clip, scaling with resolution and duration. A 1080p, 30-second clip costs significantly more, reflecting its high-fidelity output and computational demands.
No. Seedance 2.5 R2V is a closed, proprietary model developed by ByteDance. It is not open source, so self-hosting or fine-tuning is not possible. Access is via API or platform integration only.
Yes. Seedance 2.5 R2V generates synchronized audio tracks natively, including ambient sound, music, and voiceover elements, enhancing the cinematic quality of the output.
It supports 480p, 720p, and 1080p resolutions, with aspect ratios including 21:9, 16:9, 4:3, 1:1, 3:4, and 9:16, making it suitable for diverse platforms from social media to cinema.
Seedance 2.5 R2V offers superior cinematic quality, longer native duration, and multimodal reference support, while Wan 2.7 Enhanced is open-weight and more cost-effective for simpler use cases without audio or extended scenes.
Yes. Venice runs Seedance 2.5 R2V under an anonymized privacy tier — your prompts and reference materials are not stored or used for training, ensuring private, permissionless access.
Run Seedance 2.5 R2V privately
No prompt logging. No data used for training.