Wan 3.0 Prime generates video from a prompt and optional reference media.
- Tool calling
UnsupportedTool calling · Unsupported
- Reasoning
UnsupportedReasoning · Unsupported
- Vision
SupportedVision · Supported
- Streaming
UnsupportedStreaming · Unsupported
- JSON mode
UnsupportedJSON mode · Unsupported
USD per second of reference and output video combined
- 480P
- $0.04896
- 720P
- $0.1008
- 1080P
- $0.2016
- Precision
- Official precision
- Price checked
Model availability and prices
Overview
Create a video from text, guide it with a first or last frame, or supply visual and audio references. The supplier delivers an MP4 after the asynchronous job completes.
- Version
- wan3.0-video-prime
- Input / output
- text, image, video, audio / video
- Endpoint
- /v1/videos, /v1/videos/{video_id}, /v1/videos/{video_id}/content
API protocols and integration
Authorization: Bearer ITERACOMPUTE_API_KEY
POST /v1/videosGET /v1/videos/{video_id}GET /v1/videos/{video_id}/content
Submit once, poll the returned ID, then download the completed video. Billing uses reference video seconds plus generated seconds.
LLMs.txt
# wan/wan3.0-video-prime
Revision: fc7ebd1cf1f3550ee90a98afb4228ae12aa7887001ce81b456054f377375744b
Authorization: Bearer ITERACOMPUTE_API_KEY
POST https://api.iteracompute.com/v1/videos
GET https://api.iteracompute.com/v1/videos/{video_id}
GET https://api.iteracompute.com/v1/videos/{video_id}/content
- prompt · string · Description of the video. Required when no media is supplied.
- media · array · Ordered assets, each with type and url. First/last frames form a separate input mode.
- resolution · enum · Output resolution. · ["480P","720P","1080P"] · default="1080P"
- ratio · enum · Output aspect ratio. · ["adaptive","21:9","16:9","4:3","1:1","3:4","9:16"] · default="adaptive"
- duration · integer · 2–30 seconds or −1 for automatic duration. Reference video and output together may use up to 30 seconds. · [-1, 30] · default=5
- seed · integer · Random seed, −1 selects a random value. · [-1, 2147483647] · default=-1
- audio · boolean · Generate a soundtrack. · default=true
- watermark · boolean · Add the supplier watermark. · default=false
- prompt_extend · boolean · Prompt rewriting, required for file or link references. · default=true
Install requests and set ITERACOMPUTE_API_KEY. Submit once and save the task ID. Poll and download with the same key. If the wait expires, continue polling the printed ID.
import json
import time
import os
from pathlib import Path
import requests
base = "https://api.iteracompute.com/v1"
headers = {"Authorization": "Bearer " + os.environ["ITERACOMPUTE_API_KEY"]}
payload = json.loads("{\n \"model\": \"wan/wan3.0-video-prime\",\n \"prompt\": \"A blue paper boat on a calm lake\",\n \"resolution\": \"480P\",\n \"duration\": 2,\n \"audio\": false\n}")
r = requests.post(base + "/videos", headers=headers, json=payload, timeout=60)
r.raise_for_status()
job = r.json()["id"]
print("Video ID:", job, flush=True)
path = base + "/videos/" + job
deadline = time.monotonic() + 900
while time.monotonic() < deadline:
r = requests.get(path, headers=headers, timeout=30)
r.raise_for_status()
status = r.json()["status"]
if status == "completed":
r = requests.get(path + "/content", headers=headers, timeout=120)
r.raise_for_status()
Path("result.mp4").write_bytes(r.content)
break
if status in ("failed", "cancelled"):
raise RuntimeError(status)
time.sleep(5)
else:
raise TimeoutError(job)
https://www.alibabacloud.com/help/en/model-studio/wan3-video-generation-api-reference/API parameters
- Submit once, poll the returned ID, then download the completed video. Billing uses reference video seconds plus generated seconds.
These parameters and ranges describe the current route. Account and context limits also apply. Parameter acceptance does not establish model capability.
- Tool calling
- Unsupported
- Reasoning
- Unsupported
- Vision
- Supported
- Streaming
- Unsupported
- JSON mode
- Unsupported
Parameters and values
Click a parameter to copy its name; scroll the table horizontally to see all columns. Usage notes identify protocol differences; identical names do not guarantee cross-endpoint support.
| Parameter | Type | Acceptance | Range | Usage / protocol |
|---|---|---|---|---|
| integer | Accepted | [-1, 2147483647] | Random seed, −1 selects a random value. Default -1. | |
| boolean | Accepted | — | Generate a soundtrack. Default true. | |
| array | Accepted | — | Ordered assets, each with type and url. First/last frames form a separate input mode. | |
| enum | Accepted | ["adaptive","21:9","16:9","4:3","1:1","3:4","9:16"] | Output aspect ratio. Default "adaptive". | |
| string | Accepted | — | Description of the video. Required when no media is supplied. | |
| integer | Accepted | [-1, 30] | 2–30 seconds or −1 for automatic duration. Reference video and output together may use up to 30 seconds. Default 5. | |
| boolean | Accepted | — | Add the supplier watermark. Default false. | |
| enum | Accepted | ["480P","720P","1080P"] | Output resolution. Default "1080P". | |
| boolean | Accepted | — | Prompt rewriting, required for file or link references. Default true. |
Current configuration example
POST /v1/videos
{
"model": "wan/wan3.0-video-prime",
"prompt": "A blue paper boat on a calm lake",
"resolution": "480P",
"duration": 2,
"audio": false
}Parameter revision:fc7ebd1cf1f3
API
Install requests and set ITERACOMPUTE_API_KEY. Submit once and save the task ID. Poll and download with the same key. If the wait expires, continue polling the printed ID.
import json
import time
import os
from pathlib import Path
import requests
base = "https://api.iteracompute.com/v1"
headers = {"Authorization": "Bearer " + os.environ["ITERACOMPUTE_API_KEY"]}
payload = json.loads("{\n \"model\": \"wan/wan3.0-video-prime\",\n \"prompt\": \"A blue paper boat on a calm lake\",\n \"resolution\": \"480P\",\n \"duration\": 2,\n \"audio\": false\n}")
r = requests.post(base + "/videos", headers=headers, json=payload, timeout=60)
r.raise_for_status()
job = r.json()["id"]
print("Video ID:", job, flush=True)
path = base + "/videos/" + job
deadline = time.monotonic() + 900
while time.monotonic() < deadline:
r = requests.get(path, headers=headers, timeout=30)
r.raise_for_status()
status = r.json()["status"]
if status == "completed":
r = requests.get(path + "/content", headers=headers, timeout=120)
r.raise_for_status()
Path("result.mp4").write_bytes(r.content)
break
if status in ("failed", "cancelled"):
raise RuntimeError(status)
time.sleep(5)
else:
raise TimeoutError(job)Pricing
USD per second of reference and output video combined
- Model ID
wan/wan3.0-video-prime- Verification scope and sources
View verification scope and sources
2026-11-21 · Singapore tests verified 480P, 720P and 1080P, the 30-second maximum, combined reference and output duration charges, and account isolation for polling and downloads.
Usage limits
- Save the result within 24 hours. A reference video contributes to both the duration limit and the bill.
- The supplier supports 480P, 720P and 1080P. Requested output length is 2–30 seconds. With a reference video the combined duration is limited to 30 seconds.
Use cases
Integration tools and guides
FAQ
How do I get started?
How is usage billed?
New arrivals
Explore the models selected for New Arrivals, with API pricing, supported capabilities and integration guides.
Qwen3.7-Plus
qwen/qwen3.7-plus
Qwen3.7-Plus returns text from text and visual input.
Launched on IteraCompute
- Input
- $0.384
- Output
- $1.536
USD / 1M tokens
Wan 3.0
wan/wan3.0-video
Wan 3.0 generates video from a prompt and optional reference media.
Launched on IteraCompute
Qwen-Image-3.0-Pro
qwen/qwen-image-3.0-pro
Generate and edit images with Qwen-Image.
Launched on IteraCompute