Midjourney Video API
Midjourney 视频模型，支持根据起始帧、可选结束帧与提示词生成固定 5 秒视频，并可从已完成任务继续延长 5 秒；支持 SD、HD 与每次 1、2、4 个结果。
What is Midjourney Video
Midjourney Video is Midjourney's first video model (V1), released on June 18, 2025. Per the official docs it is image-to-video only: one image serves as the start frame with an optional text prompt. Official specs are fixed 5-second clips, SD around 480p and HD 720p, motion low (default) / high, plus video parameters such as raw, loop, and end — with no audio output.
Roko exposes the model through the /v1/videos/tasks async task API over the TTAPI conversion channel, with the video-task endpoint only. The current catalog offers three routes: image-to-video (1 image), start-end-to-video (2 images: start frame + end frame), and video-extend (+5 seconds from a completed task, inheriting its resolution, extend index 0–3, animate mode auto/manual).
Pricing follows the Roko production catalog: fixed per-use billing at $0.0825 per SD result and $0.3025 per HD result, with 1, 2, or 4 results per call. The extend route inherits the source task's resolution and is billed at the matching per-result tier; cost is estimated on submit and settled on actual usage at completion. Midjourney's official subscription billing differs and the two are not interchangeable.
Model pricing
Pricing follows the current production configuration and the billing dimensions supported by each route.
| Route | Price | Billing rule |
|---|---|---|
| SD | 每个结果 $0.0825/request | 按结果数固定计费 |
| HD | 每个结果 $0.3025/request | 按结果数固定计费 |
Cost examples
Examples recomputed from Roko catalog per-result prices: SD $0.0825, HD $0.3025, billed per use, estimated on submit and settled on actual usage at completion; extensions inherit the source resolution tier.
1 result × $0.0825 = $0.0825
The lowest-cost tier at the default resolution — test one clip before scaling up.
4 results × $0.0825 = $0.33
Four candidates from one start frame in a single call; the common tier for batched stylized clips.
1 result × $0.3025 = $0.3025
A single 720p-tier test clip with higher detail and sharpness.
4 results × $0.3025 = $1.21
HD batch selection; extensions inherit the HD tier at $0.3025 per result.
Model information
Only provider-documented routes, capabilities, and limits implemented by this gateway are listed.
midjourney-video
- Limits
- Images 1
- Duration
- 5s
midjourney-video
- Limits
- Images 2
- Duration
- 5s
midjourney-video
- Limits
- —
- Duration
- 5s
Key capabilities
Image-to-video with start/end frame control
image-to-video animates a still from 1 start-frame image; start-end-to-video takes 2 images (start frame + end frame) to control the shot's opening and closing transition. Text prompts are optional on both routes.
Fixed 5 seconds × extend route
Each clip is fixed at 5 seconds; video-extend continues from a completed task with another 5 seconds (extend index 0–3, animate auto/manual). The official site extends by about +4s while this channel uses +5s — defer to task results.
SD/HD resolution tiers
Defaults to SD, switchable to HD; official claims put SD around 480p and HD at 720p. Each result costs $0.0825 (SD) or $0.3025 (HD), and extensions inherit the source resolution tier.
motion low/high
motion low (official default) keeps movement restrained and cheaper, suited to subtle motion; motion high produces larger movement for visible camera moves and dynamic shots.
Batch 1/2/4 results
Each call can return 1, 2, or 4 results, billed per use: 4 SD results cost $0.33 and 4 HD results $1.21 — ideal for generating candidates from one start frame and picking the best.
Main parameters
Parameters come from structured model configuration; unsupported or unconfigured fields are omitted.
modelstringSubmodel ID
Default:midjourney-video / midjourney-video / midjourney-videopromptstringMidjourney 视频模型，支持根据起始帧、可选结束帧与提示词生成固定 5 秒视频，并可从已完成任务继续延长 5 秒；支持 SD、HD 与每次 1、2、4 个结果。
Default:RequireddurationintegerOutput video seconds
Default:5squalityenumCurrent billing
Default:sd / hdnintegerCapability
Default:1 / 2 / 4motionenumCapability
Default:low / highimage_urlsarrayImage to video: 1 · Start/end frames to video: 2
Default:Varies by routeCapabilities and usage guidance
Key features
- Image to video · midjourney-video
- Start/end frames to video · midjourney-video
- Video extension · midjourney-video
- sd / hd output
- 5s Duration
Limits
- Image to video: Images 1
- Start/end frames to video: Images 2
- Capability source: https://docs.ttapi.io/api/cn/videoapi/submit
Limitations
- Image-to-video only: there is no pure text-to-video route upstream — a start-frame image is required (text prompt optional). Catalog input modalities are text and image.
- Clips are fixed at 5 seconds; each extension adds +5 seconds (extend index 0–3). The official site extends by about +4s, up to 4 times, capped around 21s — the channel difference defers to task results.
- No audio output (official upstream claim).
- TTAPI is a conversion channel: the authoritative original specs (raw, loop, end, and other parameters) live in Midjourney's official docs; the parameters actually callable on Roko follow the catalog and task results.
- Output links expire — transfer files to your own storage before the links lapse.
How it works
- POST /v1/videos/tasks with model set to midjourney-video, attaching 1 start-frame image (image-to-video) or 2 images — start frame + end frame (start-end-to-video) — with an optional text prompt. The API returns a task ID and a price estimate.
- The task enters queueing and generation; a successful submission does not guarantee a successful generation, and billing is settled on actual usage when the task completes.
- Poll GET /v1/tasks/{task_id} until the status reaches a succeeded or failed terminal state.
- On success, read the video URL from the task result; Roko output links expire, so transfer the files to your own storage promptly. For longer footage, run video-extend on the completed task for another 5 seconds.
- Failed tasks are settled and refunded per platform rules, with the terminal task state as authoritative; the extend route must reference a completed task.
Use cases
Bring Midjourney-style stills to life
Use existing Midjourney illustrations, concept art, or character sheets as start frames to generate 5-second motion, with motion low/high dialing the intensity while keeping the visual style consistent.
Start-end frame transition shorts
Feed two style images as start and end frames via start-end-to-video to produce 5-second natural transitions — ideal for intros, scene changes, and stylized short narratives.
Low-cost batched stylized clips
At $0.0825 per SD result ($0.33 for a batch of 4), it fits social-media content pipelines: generate multiple candidates at once, pick the best, then extend the winner by 5 seconds.
How it compares
Comparisons are based on current catalog routes and pricing structures for selection guidance, not measured quality verdicts.
vs grok-imagine-video-1.5
Both run over the TTAPI per-use channel with async tasks. Midjourney Video must start from an image (image-to-video only), while Grok supports pure text-to-video. Their per-use price structures differ — choose by input modality and price tier.
vs seedance-2.0-mini
Midjourney Video starts at $0.0825 per 5-second use, seedance-2.0-mini at $0.0209 per second: per-use suits fixed 5-second batch-selection workflows, per-second suits free durations — two trade-offs in the lightweight tier.
vs veo-3.1
Midjourney Video is a stylized image-to-video + extend route with no audio, billed per use; Veo 3.1 offers native audio, 4K, and per-second billing. Pick Midjourney for style continuity, Veo for sound and high-resolution delivery.
FAQ
What is the model ID and which endpoint does it use?
The model ID is midjourney-video. Roko exposes it via the /v1/videos/tasks async API with the video-task endpoint only, over the TTAPI conversion channel. The three routes are image-to-video, start-end-to-video, and video-extend.
How is pricing calculated?
Billing is fixed per use and settled by result count: $0.0825 per SD result and $0.3025 per HD result, with 1, 2, or 4 results per call (e.g., 4 SD results $0.33, 4 HD results $1.21). An estimate is given on submit and actual usage is settled at completion; extensions inherit the source resolution and bill at the matching per-result tier.
Does it support pure text-to-video?
No. The official upstream position is image-to-video only: one image serves as the start frame with an optional text prompt. The start-end-to-video route additionally accepts a second image as the end frame.
How do duration and extension work?
Clips are fixed at 5 seconds. video-extend continues from a completed task with another 5 seconds, extend index 0–3, animate mode auto/manual, inheriting the source resolution. Note: the official Midjourney site extends by about +4s, up to 4 times, capped around 21s, while the Roko/TTAPI channel uses +5-second steps — actual behavior defers to task results.
What is the difference between SD and HD?
SD is the default tier and HD is higher. Official claims put SD around 480p and HD at 720p. Pricing is $0.0825 per SD result and $0.3025 per HD result; extensions inherit the source task's resolution tier and bill at the matching per-result price.
Do the generated videos have audio?
No. The official upstream docs state the V1 video model has no audio output; add music or voice-over in post-production if needed.
Why can't I get the video immediately?
The model runs as an async task: submission returns a task ID and price estimate first, then you poll GET /v1/tasks/{task_id} until a succeeded or failed terminal state and read the video URL from the result. Generation time depends on queue and load.
What should I know about commercial use?
Make sure you hold the rights or licenses to input images and that content does not infringe others' rights. Commercial use of generated content must comply with Midjourney's terms of service and Roko's platform terms — check the latest versions of both and review the on-site commercial-use guidance.
Sources
Upstream capability claims come from Midjourney's official docs; Roko connects through the TTAPI conversion channel, and callable parameters, routes, and prices come from the current production catalog. Verified on 2026-10-13.
- Midjourney docs: Video Official: V1 video model, image-to-video, 5s clips, motion low/high, SD/HD, batch, no audio
- TTAPI: Midjourney Video API Conversion channel API used by the Roko catalog route
- Roko API video task docs Async submit, task query, and billing lifecycle on Roko
- Roko API model catalog Current routes, prices, and generation config
Models in this family
Explore other available variants from the same model family.
Midjourney Video
Other video models
Explore more models available through the unified asynchronous video task API.
