Wan 2.6
What is Wan 2.6?
Wan 2.6 is Alibaba's December 2025 upgrade to the Wan video line. It exposes three modes on Unifically: text-to-video, image-to-video, and reference-to-video. The reference-to-video mode accepts up to five references (a mix of images and videos, with a cap of three videos) mapped to character names in the prompt. Multi-shot generation chains 2–8 segments with per-shot prompts and durations in a single call. Output runs at 720p or 1080p, 2 to 15 seconds, with auto-generated audio on by default.
Key features of Wan 2.6
Four features cover what changes versus Wan 2.5.
Reference-to-video for character consistency
Pass up to five references (images or videos, max three videos). Each reference becomes a named character in the prompt (`character1`, `character2`). Useful for ads with recurring presenters or product mascots.
Multi-shot in one call
Pass `multi_prompt` as an array of `{prompt, duration}` segments. The model chains 2–8 shots, picks shot type, and totals the duration. Useful for narrative ads with a clear arc.
2 to 15 second clips
Wider duration range than Wan 2.5. Reference-to-video caps at 10 seconds; the other two modes go to 15 seconds.
Audio in the same call
Auto-generated audio on by default. Pass `audio_url` for custom sync, or set `audio: false` for silent output. Same audio surface as Wan 2.5 with the longer clip range.
Best for
Long-form clips up to 15 seconds
Single-clip duration extends to 15 seconds in text-to-video and image-to-video. Useful for narrative ads and explainer beats.
Multi-shot narratives
2–8 segments per call with per-segment prompts. Useful for ads with a clear setup-beat-payoff arc.
Reference-to-video character continuity
Up to five references mapped to character names in the prompt. Useful for recurring presenters and product mascots.
Custom audio sync
Upload your own track when the audio mix is fixed and visuals need to match the beats.
Negative-prompt control
Suppress unwanted content in text-to-video and image-to-video for brand-safe output.
Established production pipelines
Mature parameter surface for workflows already running Wan 2.6.
Variants
Wan 2.6 has two callable variants on the same API. Each one has its own model ID and price.
Wan 2.6 (standard)
The standard model. Use it for the final pass on hero clips, paid placements, and anything where peak quality matters more than throughput.
Wan 2.6 Flash
Faster queue and lower price per clip on the same parameter surface. Use it for prompt sweeps, draft passes, and high-volume batches before locking the keepers on the standard model.
API examples
Call Wan 2.6 from any language by POSTing to /v1/tasks. Full parameter docs live at docs.unifically.com/models/video/alibaba/wan-2.6-video.
curl -X POST https://api.unifically.com/v1/tasks \
-H "Content-Type: application/json" \
-H "Authorization: Bearer YOUR_API_KEY" \
-d '{
"model": "alibaba/wan-2.6-video",
"input": {
"prompt": "A cinematic shot of a red fox walking through a misty forest at dawn",
"mode": "t2v",
"resolution": "720p",
"duration": 5
}
}'
Successful submission returns a task_id. Poll GET /v1/tasks/<task_id> or set a callback_url on the request to receive the finished result.
FAQs
People also ask
Wan 2.6 is Alibaba's December 2025 video model. It supports text-to-video, image-to-video, and reference-to-video with up to five references (mix of images and videos, max three videos). Multi-shot mode chains 2–8 segments in one call. Output runs at 720p or 1080p, 2 to 15 seconds, with optional generated or custom audio.
Three. text-to-video (t2v) from a prompt, image-to-video (i2v) from a start_image_url, and reference-to-video (r2v) where you supply up to five reference files mapped to character names in the prompt.
Yes. Audio generation is on by default. Set audio: false for silent output, or pass an audio_url to drive audio-video sync to a fixed track.
2 to 15 seconds in text-to-video and image-to-video. Reference-to-video is capped at 2 to 10 seconds. Multi-shot duration is the sum of all segment durations.
Negative prompts apply to text-to-video and image-to-video. They are ignored in reference-to-video.
Use Wan 2.6 for established workflows and predictable single-reference scenes. Use Wan 2.7 when you need last-frame control, video continuation, lip-sync from a driving audio track, or up to five video references.
Same parameter shape and same modes. Wan 2.6 Flash runs on a faster queue at a lower price. Use Flash for throughput batches; use the standard Wan 2.6 for the keepers.
