Q3 reference-to-video model supporting image and text subjects, camera changes and synchronized sound. Distinct from the Q3 Pro selector used by text and image routes.
Submit a native task and poll /ent/v2/tasks/{id}/creations, or provide a signed callback URL. JSON task envelopes are not evidence of LLM structured-output support.Source for Request mode (opens in a new tab)
Task callbacks
Processing, success and failure events. Terminal callback delivery is retried up to three times. HMAC-SHA256 signature uses the provider-documented canonical string, signed headers, date and nonce.Source for Task callbacks (opens in a new tab)
Default five concurrent tasks per organization, shared across API keys; excess tasks queue. Purchased plan and system load can change effective concurrency.Source for Request quotas (opens in a new tab)
Usage metering
Generated-video seconds, with prices dependent on route, resolution, account market and eligible off-peak mode. Not a text-token or per-question tariff.Source for Usage metering (opens in a new tab)
Regional credit values are separate provider prices, not a foreign-exchange conversion.
Off-peak generation
Off-peak is false by default. Provider promises completion within 48 hours or automatic cancellation with credit refund. Manual cancellation is available before disallowed task states; the cancellation API shows scheduled tasks cannot be cancelled. Off-peak scheduling is not evidence of a discounted batch API.Source for Off-peak generation (opens in a new tab)
Reusable subjects
Reference route supports image/text subjects, @name references, optional saved server_id and temporary images. q2-pro video-subject examples in the shared schema do not establish video input for Q3.Source for Reusable subjects (opens in a new tab)
Access & pricing
Prices belong to an endpoint and billing tier. A provider listing does not guarantee access for every account or region.
Off-peak is false by default. Provider promises completion within 48 hours or automatic cancellation with credit refund. Manual cancellation is available before disallowed task states; the cancellation API shows scheduled tasks cannot be cancelled. Off-peak scheduling is not evidence of a discounted batch API.Source for Off-peak generation (opens in a new tab)
Off-peak pricing conditions
International reference pricing states half price rounded to the nearest integer credit. Half-credit tie behavior is not specified; exact odd-credit discounted rates are withheld rather than copied from mainland tables.Source for Off-peak pricing conditions (opens in a new tab)
Prices retain their published currencies and billing units below.
Published billing units · original currency · observed 14 Oct 2026
Category
Usage
Rate
Unit
Video generation
540p · standard1
$0.035
second of output video
Video generation
720p · standard2
$0.06
second of output video
Off-peak video generation
720p · off-peak3
$0.03
second of output video
Video generation
1080p · standard4
$0.075
second of output video
Rate conditions
7 credits per output second × USD 0.005 per credit, calculated from the published regional schedule. Applies to viduq3 on POST https://api.vidu.com/ent/v2/reference2video. Base generation price; prompt expansion, subject creation and other add-ons are separate. Promotions and taxes are excluded.
12 credits per output second × USD 0.005 per credit, calculated from the published regional schedule. Applies to viduq3 on POST https://api.vidu.com/ent/v2/reference2video. Base generation price; prompt expansion, subject creation and other add-ons are separate. Promotions and taxes are excluded.
6 credits per output second × USD 0.005 per credit, calculated from the published regional schedule. Applies to viduq3 on POST https://api.vidu.com/ent/v2/reference2video. Base generation price; prompt expansion, subject creation and other add-ons are separate. Promotions and taxes are excluded. Half-price calculation has an integer result at this resolution; ambiguous half-credit ties are not quoted.
15 credits per output second × USD 0.005 per credit, calculated from the published regional schedule. Applies to viduq3 on POST https://api.vidu.com/ent/v2/reference2video. Base generation price; prompt expansion, subject creation and other add-ons are separate. Promotions and taxes are excluded.
International privacy policy states personal data storage in Singapore and retention for legal/operational purposes. It does not establish inference residency, zero retention or a no-training guarantee. Creation links last 24 hours.
Mainland fields expose watermark, position, image URL and metadata. Some fields state true by default while nearby prose says no default watermark; set explicitly and verify.Source for Visible watermark (opens in a new tab)
Off-peak generation
Off-peak is false by default. Provider promises completion within 48 hours or automatic cancellation with credit refund. Manual cancellation is available before disallowed task states; the cancellation API shows scheduled tasks cannot be cancelled. Off-peak scheduling is not evidence of a discounted batch API.Source for Off-peak generation (opens in a new tab)
Prices retain their published currencies and billing units below.
Published billing units · original currency · observed 14 Oct 2026
Category
Usage
Rate
Unit
Video generation
540p · standard1
CN¥0.21875
second of output video
Off-peak video generation
540p · off-peak2
CN¥0.125
second of output video
Video generation
720p · standard3
CN¥0.375
second of output video
Off-peak video generation
720p · off-peak4
CN¥0.1875
second of output video
Video generation
1080p · standard5
CN¥0.46875
second of output video
Off-peak video generation
1080p · off-peak6
CN¥0.21875
second of output video
Rate conditions
7 credits per output second × CNY 0.03125 per credit, calculated from the published regional schedule. Applies to viduq3 on POST https://api.vidu.cn/ent/v2/reference2video. Base generation price; prompt expansion, subject creation and other add-ons are separate. Promotions and taxes are excluded.
4 credits per output second × CNY 0.03125 per credit, calculated from the published regional schedule. Applies to viduq3 on POST https://api.vidu.cn/ent/v2/reference2video. Base generation price; prompt expansion, subject creation and other add-ons are separate. Promotions and taxes are excluded. Uses the explicit mainland reference off-peak column, including the published 15-to-7-credit row.
12 credits per output second × CNY 0.03125 per credit, calculated from the published regional schedule. Applies to viduq3 on POST https://api.vidu.cn/ent/v2/reference2video. Base generation price; prompt expansion, subject creation and other add-ons are separate. Promotions and taxes are excluded.
6 credits per output second × CNY 0.03125 per credit, calculated from the published regional schedule. Applies to viduq3 on POST https://api.vidu.cn/ent/v2/reference2video. Base generation price; prompt expansion, subject creation and other add-ons are separate. Promotions and taxes are excluded. Uses the explicit mainland reference off-peak column, including the published 15-to-7-credit row.
15 credits per output second × CNY 0.03125 per credit, calculated from the published regional schedule. Applies to viduq3 on POST https://api.vidu.cn/ent/v2/reference2video. Base generation price; prompt expansion, subject creation and other add-ons are separate. Promotions and taxes are excluded.
7 credits per output second × CNY 0.03125 per credit, calculated from the published regional schedule. Applies to viduq3 on POST https://api.vidu.cn/ent/v2/reference2video. Base generation price; prompt expansion, subject creation and other add-ons are separate. Promotions and taxes are excluded. Uses the explicit mainland reference off-peak column, including the published 15-to-7-credit row.
Vidu activation in Model Studio is required. POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis with Bearer API key, application/json and X-DashScope-Async: enable. Query GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}.
Only successful output is billed per generated video second. No input-media tariff is published for these routes. The response usage.duration reports billable duration.Source for Output billing (opens in a new tab)
540P: 960*528, 576*1024, 1024*1024, 1024*768, 768*1024; 720P: 1280*720, 720*1280, 1280*1280, 1280*960, 960*1280; 1080P: 1920*1080, 1080*1920, 1920*1920, 1920*1440, 1440*1920Source for Output pixel presets (opens in a new tab)
540P default prose says 1024*576 but the valid-size table says 960*528 for 16:9. Preserve this conflict; do not silently substitute the text-to-video size table.
Prices retain their published currencies and billing units below.
Published billing units · original currency · observed 14 Oct 2026
Category
Usage
Rate
Unit
Video generation
540P1
CN¥0.3125
second of output video
Video generation
720P1
CN¥0.625
second of output video
Video generation
1080P1
CN¥0.78125
second of output video
Rate conditions
Published original Beijing rate for vidu/viduq3_reference2video. Promotions excluded; no free quota. Only successful output seconds are billed. The card publishes one rate per resolution, with no separately itemized audio surcharge. Do not apply native Vidu credit, per-clip or off-peak formulas to this hosted rate.
Vidu is the inference provider. The Beijing API region and 24-hour output-link lifetime do not establish inference data residency, input retention, training use or deletion policy.
Caching
Caching can reuse earlier input processing. Availability, lifetime and billing depend on the endpoint. A cache hit still generates a new response.
01 / INITIAL INPUTProcess & retain
The system processes eligible input and retains reusable state.
02 / LATER REQUESTCheck available state
Reuse depends on matching input, session state and the cache policy.
03 / CACHE HITReuse & generate
Reusing state can reduce repeated work. Any charges follow the billing rules below.
Vidu API · international · reference
viduq3
Not verified
No explicit prompt-cache or cached-input tariff in the reviewed video contract.
How it works
Not collected
No explicit prompt-cache or cached-input tariff in the reviewed video contract.
Minimum reusable input
Not collected
No explicit prompt-cache or cached-input tariff in the reviewed video contract.
Lifetime & refresh
Not collected
No explicit prompt-cache or cached-input tariff in the reviewed video contract.
How to enable it
Not collected
No explicit prompt-cache or cached-input tariff in the reviewed video contract.
Usage fields
Not collected
No explicit prompt-cache or cached-input tariff in the reviewed video contract.
Billing rules
Not collected
No explicit prompt-cache or cached-input tariff in the reviewed video contract.
Limits & data boundaries
Not collected
No explicit prompt-cache or cached-input tariff in the reviewed video contract.
No verified cache rate for this endpoint. An empty rate is not a zero charge.
Caching reviewed 14 Oct 2026
Vidu API · mainland China · reference
viduq3
Not verified
No explicit prompt-cache or cached-input tariff in the reviewed video contract.
How it works
Not collected
No explicit prompt-cache or cached-input tariff in the reviewed video contract.
Minimum reusable input
Not collected
No explicit prompt-cache or cached-input tariff in the reviewed video contract.
Lifetime & refresh
Not collected
No explicit prompt-cache or cached-input tariff in the reviewed video contract.
How to enable it
Not collected
No explicit prompt-cache or cached-input tariff in the reviewed video contract.
Usage fields
Not collected
No explicit prompt-cache or cached-input tariff in the reviewed video contract.
Billing rules
Not collected
No explicit prompt-cache or cached-input tariff in the reviewed video contract.
Limits & data boundaries
Not collected
No explicit prompt-cache or cached-input tariff in the reviewed video contract.
No verified cache rate for this endpoint. An empty rate is not a zero charge.
Caching reviewed 14 Oct 2026
Alibaba Model Studio · Beijing · Reference to video
No benchmark observations have been added for this model. We do not turn a vendor description into a performance score.
Latency and throughput depend on the host, request size, reasoning settings and load. There is no measured speed or global usage figure in this record.
The record behind the model
Why it is included
Qualified closed-weight classification inference from a separately named Vidu native hosted model and operator terms reserving proprietary service rights. Public API clients and efficiency libraries do not establish downloadable weights. Exact checkpoint lineage remains unverified.
