English · 繁體中文

Developer API

REST · JSON · MCP · version v1 · Changelog · Service status

Create videos from your own code, or let an AI agent do it through our MCP server. Both are for business accounts: they use the same sign-in as the website, with a separate business balance. Prices and balances are in US dollars. Business accounts prepay in USDC, 1 USDC for $1 of balance, and pay 40% less than the website.

At a glance

FormatBusiness priceTop-up of 1,000 USDC or more
Video720p portrait (736 × 1280), 3, 5, 7 seconds, up to 3 reference images$0.10 per second$0.08 per second
Image720p portrait (736 × 1280), up to 3 reference images$0.05 each$0.04 each

Prices in US dollars. The API and MCP are for business accounts, which prepay in USDC: 1 USDC for $1 of balance, minimum top-up 1 USDC, 40% off the website price (about $0.167 per second). A single top-up of 1,000 USDC or more adds 25% extra balance. Contact us for a business account (sales@mytype.video).

Default limits: 60 reads and 10 writes per minute per token; 2 jobs running and 20 waiting per account. Volume pricing and higher limits on request.

Quick start

  1. Create a token under Account settings › API & MCP (you sign up or log in first) and copy it: it is shown once.
  2. Send it as a bearer token on every request to https://mytype.video/api/v1.
curl https://mytype.video/api/v1/session \
  -H "Authorization: Bearer $MYTYPE_TOKEN"

Tokens

  • Any account with a verified email can create a token. Each account can have up to five active tokens, for example one for testing and one for production. To replace a token without downtime, create the new one, switch your servers to it, then revoke the old one. A sixth returns 409 TOKEN_LIMIT_REACHED.
  • A token expires after 30, 90 or 365 days. We store only a hash, so a lost token cannot be recovered: revoke it and create a new one.
  • Access. A read token can read your account, wallet, templates, media and jobs, and ask for quotes. Starting videos and uploading images needs generate. A request outside the token’s access returns 403 SCOPE_REQUIRED.
  • IP allowlist (optional). In Account settings › API & MCP you can limit a token to up to 20 IPv4/IPv6 addresses or CIDR ranges. Tokens have no allowlist unless you set one. A request from another address returns 403 IP_NOT_ALLOWED. Through MCP the address is the one your MCP client connects from.
  • Usage. Account settings shows what each token spent this month, in US dollars. Tokens have no spending limit: what a token can spend is your prepaid balance.
  • Tokens always act as a normal user account. Bans and email verification are checked on every request.

Prices and billing

Prices are in US dollars. At a glance lists them.

  • Business accounts only. API and MCP jobs are paid from a business balance, so an account needs a business balance to use them. Contact us to open one; we can add trial credit for a first test.
  • Business price. For business accounts that prepay in USDC: 1 USDC adds $1 to your balance. The minimum top-up is 1 USDC. A single top-up of 1,000 USDC or more adds 25% extra balance ($0.08 per second of video, $0.04 per image); separate top-ups are not added together. Your account settings show Business account, and GET /wallet returns account.type "business". Contact us to open one.
  • Separate balances. The API and MCP spend only the business balance (USDC top-ups and trial credit). Points bought or received on the website are a separate balance for the website, and neither pays for the other. GET /wallet shows both in balanceByPool.
  • Every way of starting a video has a quote twin that validates the same body and returns its cost in US dollars (amountUsd) without starting or charging anything.
  • Internally the API also counts in points, 10 per second of video and 5 per image, and older fields such as points stay. account.pointsPerUsd in GET /wallet converts them: at the business list rate, 1 point is $0.01.

How a job is charged

  1. Quote. Free. Nothing is reserved and no job is created.
  2. Accepted. The quoted amount is moved from your balance to a hold. If your balance is too small the request returns 402 INSUFFICIENT_BALANCE and creates nothing; top up and send the same requestId again.
  3. Completed. The hold is charged. You pay only for jobs that finish.
  4. Failed. The hold is returned in full: on a technical failure, when a job has not finished 20 minutes after it was accepted, or when its result cannot be published within 15 minutes after rendering.
  • GET /wallet returns your balance in US dollars (balanceUsd), your account type and price (account), and every hold, charge and return.
  • Paying for API use. Business customers prepay in USDC, at least 1 USDC at a time. Once your account is approved, Account settings shows your own deposit address on Base and the payment is added to your balance automatically. Contact us or write to sales@mytype.video to get started. A prepaid balance does not expire while the account is in use, and there is no automatic top-up: keep an eye on your balance. The API terms of use have the details.
  • Planning a high monthly volume? Contact us to discuss pricing and limits.

Upload, generate, download

Upload a PNG, JPEG or WebP image up to 2 MiB. You must confirm that you have the right to use it (rightsAttested: true), exactly as on the website; NSFW content also requires adultSubjectsVerified: true. Everyone depicted must be an adult.

# 1. Reserve the upload. sha256 is the hex SHA-256 of the file, bytes its size.
curl -X POST https://mytype.video/api/v1/assets \
  -H "Authorization: Bearer $MYTYPE_TOKEN" -H "Content-Type: application/json" \
  -d '{"filename":"photo.jpg","title":"My photo","contentType":"image/jpeg",
       "bytes":183204,"sha256":"<sha256 of the file>",
       "contentRating":"sfw","style":"realism","rightsAttested":true}'
# -> {"uploadId":"upl_…","assetVersionId":"av_…","upload":{"url":"…","headers":{…},"expiresAt":"…"}}

# 2. PUT the file bytes to upload.url with every header in upload.headers, before upload.expiresAt.

# 3. Ask for validation, then read the image until it is available.
curl -X POST https://mytype.video/api/v1/assets/uploads/<uploadId>/complete -H "Authorization: Bearer $MYTYPE_TOKEN"
curl https://mytype.video/api/v1/assets/<assetVersionId> -H "Authorization: Bearer $MYTYPE_TOKEN"

Then quote, start and follow the video. The examples are complete programs; replace <assetVersionId> with your uploaded image.

# Price it first (read scope, nothing is charged).
curl -X POST https://mytype.video/api/v1/generations/direct/quote \
  -H "Authorization: Bearer $MYTYPE_TOKEN" -H "Content-Type: application/json" \
  -d '{"mode":"ref2va","requestId":"<a new UUID>","resolution":"736p","duration":5,"seed":42,
       "prompt":"The person in <Picture 1> turns and smiles at the camera.",
       "contentRating":"sfw","style":"realism","motion":"idle",
       "referenceAssetVersionIds":["<assetVersionId>"]}'
# -> {"amountUsd":0.5,"currency":"USD","accountType":"business","points":50,"pointsPerSecond":10}

# Start it: the same body to /generations/direct. Reuse the same requestId when you retry.
# -> 202 {"jobId":"job_piper_…","status":"accepted"}

# Read the job every 5 to 10 seconds until it is completed or failed.
curl https://mytype.video/api/v1/generations/<jobId> -H "Authorization: Bearer $MYTYPE_TOKEN"

A started video cannot be cancelled. Media links are signed and expire (expiresAt); read the job or asset again for a fresh link. Finished videos are kept with your account and have no automatic expiry.

Video and image formats

  • Video. 720p portrait (736 × 1280), 3, 5 or 7 seconds, from a prompt and up to three reference images. Send "resolution":"736p" with the request. Priced per second (see Prices and billing).
  • First and last frame. POST /generations/frame-to-frame makes a video that moves from one of your images to another, in the same format and lengths.
  • Images. POST /generations/image makes a 720p portrait picture (736 × 1280) from a prompt and up to three reference images. Priced per image.
  • GET /capabilities lists what your account can use right now.

MCP server

AI agents can use the same account through our MCP server instead of calling the REST API. It speaks Streamable HTTP at https://mcp.mytype.video/mcp and takes the same token as a bearer header. Reading and quoting need read access; uploading and generating need generate. Prices, balance, limits and content rules are the same as for the API.

  1. Create a token. Give an agent its own token, so you can see what it spends and revoke it on its own.
  2. Add the server to your client. Pick yours:
claude mcp add --transport http mytype https://mcp.mytype.video/mcp \
  --header "Authorization: Bearer $MYTYPE_TOKEN"

Any other client that supports Streamable HTTP with a custom Authorization header works the same way. Or let your coding agent do it: paste this message into Claude Code, Cursor, Codex or another agent that can edit its own settings.

Connect the MyType.video MCP server for me.
Server: https://mcp.mytype.video/mcp (Streamable HTTP).
Authentication: send my token as the header "Authorization: Bearer <token>". The token is in the environment variable MYTYPE_TOKEN; never print it or write it into a file that is committed.
After connecting, call get_capabilities and get_video_workflow, tell me my balance in US dollars, and quote a 5 second 720p video before you start anything that costs money.
Documentation: https://mytype.video/developers#mcp

The server does not offer an OAuth sign-in yet, so apps that connect only through a sign-in (such as the Claude and ChatGPT apps) cannot add it as a connector today.

  • Ask the agent to call get_capabilities and get_video_workflow first. They return the current formats and the recipes for multi-clip videos.
  • quote_generation prices a request without charging; submit_generation starts it and get_generation reads the result. Both take {"request": {"kind": "direct", "body": {…}}}, where body is the same JSON as the REST request. For 720p video the body must include "resolution":"736p".

Job status

Read a job with GET /generations/{jobId} every 5 to 10 seconds, or set a webhook and skip polling. Its status moves through:

  • queued: accepted and waiting for a worker (progress.phase is pending_dispatch). The first job after a quiet period can wait about 2 minutes while a worker starts.
  • running: rendering.
  • uploading: rendering is done and the result is being stored. Keep polling.
  • completed: each item has a downloadUrl. Download it promptly; read the job again for a fresh link.
  • failed: the held amount is returned. Send a new job with a new requestId.

Webhooks

Instead of polling, your account can receive a POST when each of its jobs ends.

  • Open Account settings › API & MCP and enter a Webhook address. The first save shows a signing secret once: keep it on your server. The address must be public https with a host name, on the default port. API tokens cannot change this setting.
  • One POST per job when it becomes completed or failed, for every job of the account (API and website). event is generation.completed or generation.failed; a failed job has points: 0 (nothing charged) and no items. No prompts or other inputs are sent.
  • downloadUrl expires after a few minutes: download at once, or read GET /generations/{jobId} for a fresh link.
  • Signature. X-MyType-Signature: t=<unix seconds>,v1=<hex> is HMAC-SHA256 with your secret over t + "." + raw body. Reject a t more than 5 minutes old. For 24 hours after you rotate the secret there are two v1 values, one per secret.
  • Delivery. Answer with any 2xx within 10 seconds. Otherwise we retry after 1 minute, 5 minutes, 30 minutes, 2 hours and 6 hours, then stop. Redirects are not followed. A timeout after you processed it causes a retry, so deduplicate on jobId.
  • Turning webhooks off drops deliveries that have not gone out yet. Keep polling as a fallback if a delivery can be missed.
{"event":"generation.completed","jobId":"job_piper_…","requestId":"<your requestId>",
 "status":"completed","points":30,"occurredAt":"…",
 "items":[{"assetVersionId":"av_…","kind":"video","downloadUrl":"https://…","expiresAt":"…"}]}
// Node.js: verify before trusting the body. rawBody is the exact bytes received.
import {createHmac, timingSafeEqual} from "node:crypto";
function verified(header, rawBody, secret) {
  const parts = header.split(","), t = parts.find(p => p.startsWith("t="))?.slice(2);
  if (!t || Math.abs(Date.now() / 1000 - Number(t)) > 300) return false;
  const expected = createHmac("sha256", secret).update(t + "." + rawBody).digest();
  return parts.filter(p => p.startsWith("v1=")).some(p => {
    const given = Buffer.from(p.slice(3), "hex");
    return given.length === expected.length && timingSafeEqual(given, expected);
  });
}

Low-balance alerts

Jobs stop with 402 when the balance runs out, and prepaid USDC is never charged automatically. Set an alert so you can add funds first.

  • In Account settings › API & MCP, enter a Low-balance alert amount in US dollars ($1 to $100,000). API tokens cannot change this setting.
  • When your available balance falls below it, we email the account address and, if a webhook is set, POST a balance.low event to it. Balances are checked every few minutes, not on each job.
  • One notice per drop. The alert resets once the balance is back at or above the amount, usually after a top-up. Changing the amount checks again.
  • The POST uses the same signature, headers and retries as job webhooks. X-MyType-Delivery is the eventId: deduplicate on it.
{"event":"balance.low","eventId":"balance_low_…","balanceUsd":42.5,"thresholdUsd":100,
 "currency":"USD","occurredAt":"…"}

Retries

Every request that starts a video carries a requestId (a UUID you generate). If a request times out or you are unsure it arrived, send it again with the same requestId and the same body: you get the original job back and are never charged twice. The same requestId with a different body returns 409.

Limits

  • Request rate. By default each token may make 60 reads and 10 writes per minute; quotes and POST /generations/events count as reads. Each account may make twice that across the website and all its tokens. We may set different limits for your account. Over the limit you get 429 RATE_LIMITED with a Retry-After header in seconds.
  • Jobs at a time. By default each account runs 2 jobs at once and keeps up to 20 waiting; further jobs wait in your queue. GET /queue returns the limits that apply to your account. Need more? Contact us. When your queue or the service queue is full the API returns 429 (ACCOUNT_QUEUE_FULL or QUEUE_FULL) with Retry-After. GET /capabilities returns the limits and GET /queue your current counts.
  • A rejected request creates nothing and charges nothing.

Errors

Errors are JSON: {"error": "message", "code": "CODE"}. code is present when the status alone is not enough.

StatusMeaning
400Invalid request: malformed JSON, unsupported fields or values. Nothing is created or charged.
401Missing, malformed, revoked or expired token (code TOKEN_REQUIRED when no token is sent).
402Not enough balance in the account (code INSUFFICIENT_BALANCE), or the token reached its monthly limit (code SPEND_CAP_REACHED). Nothing is charged; top up or raise the limit and send the same requestId again. The website keeps working for the same account.
403The token lacks the required scope or the route is outside the public API (code SCOPE_REQUIRED), the request address is outside the token's optional IP allowlist (code IP_NOT_ALLOWED), or the account is banned or unverified.
404The resource does not exist or is not accessible to this account.
409The request conflicts with current state: the same requestId was reused with a different body, or an upload or asset is not in a usable state (see x-error-codes).
429Request rate limit (code RATE_LIMITED) or waiting-job limit (codes ACCOUNT_QUEUE_FULL, QUEUE_FULL) reached. Nothing is created or charged; retry after the Retry-After seconds.
503New work is temporarily not admitted (codes GENERATION_PAUSED, PROTECTION_UNAVAILABLE). Nothing is charged; honor Retry-After when present.
CodeStatus and meaning
TOKEN_REQUIRED401: supply a personal API token
SCOPE_REQUIRED403: missing scope or public route permission
SPEND_CAP_REACHED402: token monthly points cap; website remains usable
RATE_LIMITED429: edge, account or token request rate; honor Retry-After
ACCOUNT_QUEUE_FULL429: account waiting-job limit
QUEUE_FULL429: service waiting-job limit
GENERATION_PAUSED503: no new admissions; accepted retries remain idempotent
PROTECTION_UNAVAILABLE503: request protection is unavailable; honor Retry-After
UPLOAD_PENDING_LIMIT429: finish pending uploads before reserving more
UPLOAD_DAILY_LIMIT429: daily reservation count or bytes reached; resets at UTC midnight
UPLOAD_STORAGE_LIMIT409: account uploaded-image storage allowance reached
UPLOAD_TRANSFER_BUSY409: concurrent PUT or exhausted transfer retries
UPLOAD_EXPIRED409: upload expired or transfer incomplete; reserve a new upload if expired
UPLOAD_CONTENT_REJECTED400: image failed the media content check; upload a different image
UPLOAD_VALIDATION_UNAVAILABLE503: image screening could not complete after retries; start a new upload
PUBLICATION_ADMIN_ONLY403: customer sharing controls are disabled; publication requires administrator approval
FRAME_ALREADY_EXISTS409: a temporary last frame already occupies this generated-video slot

Endpoints

All paths are relative to https://mytype.video/api/v1. This list is built from the live OpenAPI document, which also has the full response schemas.

GET /usageread

Read your usage in US dollars by day and by API token

GET /sessionread

Read your account identity

GET /capabilitiesread

Read supported settings, points per second and limits

GET /walletread

Read your balance (in points and US dollars), account type and activity

GET /queueread

Read your waiting/executing counts, limits and admission availability

GET /posesread

List accessible published templates

GET /poses/{poseId}read

Read one published pose with the same public fields as the catalog

GET /assetsread

List your media and accessible shared media

POST /assetsgenerate

Reserve an image upload. PUT bytes to upload.url with upload.headers before upload.expiresAt; then request completion. Adult verification is required for NSFW uploads.

Request body
FieldTypeNotes
filenamestringRequired.
titlestringRequired.
contentType"image/png" | "image/jpeg" | "image/webp"Required.
bytesintegerRequired.
sha256stringRequired.
contentRating"sfw" | "nsfw"Required.
stylestringRequired.
rightsAttestedtrueRequired.
adultSubjectsVerifiedboolean
visibility"private" | "shared"
tagsstring[]
derivedFromobject
externalVideoSourceobject

GET /assets/{assetVersionId}read

Read accessible media and refresh its signed URL

GET /assets/{assetVersionId}/reuseread

Read currently reusable template settings

POST /assets/uploads/{uploadId}/completegenerate

Request image validation. Retry completion to get 202 while processing, 200 when available, or 400 UPLOAD_CONTENT_REJECTED after a failed content check.

POST /quoteread

Validate new intent and quote points without admitting a job or reserving points

Request body: GenerationIntent
FieldTypeNotes
requestIdstring (uuid)Required.
poseIdstringRequired.
poseRevisionstring (date-time)Required.
characterAssetVersionIdstringRequired.
secondaryAssetVersionIdstring
duration3 | 5 | 7Required. Clip length in seconds. Output is 736 × 1280 (720p). 1 and 10 seconds are no longer offered.
seedintegerRequired.
motion"idle" | "transition"Required.
reuseAssetVersionIdstring
contextContextRequired.
Request body: ExtensionIntent (mode "extend")
FieldTypeNotes
mode"extend"Required.
requestIdstring (uuid)Required.
sourceFrameAssetVersionIdstringRequired.
duration3 | 5Required.
seedintegerRequired.
promptstringRequired.
audiostringRequired.

GET /generationsread

List your jobs

POST /generationsgenerate

Accept a template or Extend job and reserve points atomically

Request body: GenerationIntent
FieldTypeNotes
requestIdstring (uuid)Required.
poseIdstringRequired.
poseRevisionstring (date-time)Required.
characterAssetVersionIdstringRequired.
secondaryAssetVersionIdstring
duration3 | 5 | 7Required. Clip length in seconds. Output is 736 × 1280 (720p). 1 and 10 seconds are no longer offered.
seedintegerRequired.
motion"idle" | "transition"Required.
reuseAssetVersionIdstring
contextContextRequired.
Request body: ExtensionIntent (mode "extend")
FieldTypeNotes
mode"extend"Required.
requestIdstring (uuid)Required.
sourceFrameAssetVersionIdstringRequired.
duration3 | 5Required.
seedintegerRequired.
promptstringRequired.
audiostringRequired.

GET /generations/{jobId}read

Read your job state and available outputs

POST /generations/{jobId}/cancelgenerate

Cancellation is disabled after durable backend acceptance.

POST /generations/directgenerate

Submit a direct-prompt portrait video job

Request body: DirectGenerationIntent (mode "fl2va")
FieldTypeNotes
mode"fl2va"Required.
requestIdstring (uuid)Required.
duration3 | 5 | 7Required.
seedintegerRequired.
promptstringRequired. Full prompt passed through as written. The renderer also enforces a 2,048-token limit.
contentRating"sfw" | "nsfw"Required. NSFW inputs require NSFW output.
stylestringRequired. Rendering style. Use realism (photographic) or anime; these are the values the catalog and examples use. Other text up to 64 characters is accepted and stored with the job, but no other style is supported.
adultSubjectsVerifiedbooleanRequired true for NSFW. Confirms fictional or consenting verified-adult subjects.
firstFrameAssetVersionIdstringRequired. Owned image asset version for the opening frame.
lastFrameAssetVersionIdstringOwned image asset version for the final frame. Reuse the opening image for idle motion.
referenceAssetVersionIdstringOwned visual reference for 3-, 5- or 7-second FL2VA. Address it as <Picture 2> without a last frame, or <Picture 3> with a last frame.
referenceAssetVersionIdsstring[]Ordered owned or eligible admin-published references for 3-, 5- or 7-second FL2VA without a last frame. Address these as <Picture 2> and <Picture 3>.
resolution"736p"Portrait video resolution. Direct video is 736p (736 × 1280) only; omit the field or set it to 736p. 544p is no longer offered.
Request body: DirectGenerationIntent (mode "ref2va")
FieldTypeNotes
mode"ref2va"Required.
requestIdstring (uuid)Required.
duration3 | 5 | 7Required.
seedintegerRequired.
promptstringRequired. Full prompt passed through as written. The renderer also enforces a 2,048-token limit.
contentRating"sfw" | "nsfw"Required. NSFW inputs require NSFW output.
stylestringRequired. Rendering style. Use realism (photographic) or anime; these are the values the catalog and examples use. Other text up to 64 characters is accepted and stored with the job, but no other style is supported.
adultSubjectsVerifiedbooleanRequired true for NSFW. Confirms fictional or consenting verified-adult subjects.
referenceAssetVersionIdsstring[]Required. Owned image asset versions in Picture 1 through Picture 3 order.
motion"idle" | "transition"Required.
resolution"736p"Portrait video resolution. Direct video is 736p (736 × 1280) only; omit the field or set it to 736p. 544p is no longer offered.

POST /generations/direct/quoteread

Quote a direct-prompt portrait video job

Request body: DirectGenerationIntent (mode "fl2va")
FieldTypeNotes
mode"fl2va"Required.
requestIdstring (uuid)Required.
duration3 | 5 | 7Required.
seedintegerRequired.
promptstringRequired. Full prompt passed through as written. The renderer also enforces a 2,048-token limit.
contentRating"sfw" | "nsfw"Required. NSFW inputs require NSFW output.
stylestringRequired. Rendering style. Use realism (photographic) or anime; these are the values the catalog and examples use. Other text up to 64 characters is accepted and stored with the job, but no other style is supported.
adultSubjectsVerifiedbooleanRequired true for NSFW. Confirms fictional or consenting verified-adult subjects.
firstFrameAssetVersionIdstringRequired. Owned image asset version for the opening frame.
lastFrameAssetVersionIdstringOwned image asset version for the final frame. Reuse the opening image for idle motion.
referenceAssetVersionIdstringOwned visual reference for 3-, 5- or 7-second FL2VA. Address it as <Picture 2> without a last frame, or <Picture 3> with a last frame.
referenceAssetVersionIdsstring[]Ordered owned or eligible admin-published references for 3-, 5- or 7-second FL2VA without a last frame. Address these as <Picture 2> and <Picture 3>.
resolution"736p"Portrait video resolution. Direct video is 736p (736 × 1280) only; omit the field or set it to 736p. 544p is no longer offered.
Request body: DirectGenerationIntent (mode "ref2va")
FieldTypeNotes
mode"ref2va"Required.
requestIdstring (uuid)Required.
duration3 | 5 | 7Required.
seedintegerRequired.
promptstringRequired. Full prompt passed through as written. The renderer also enforces a 2,048-token limit.
contentRating"sfw" | "nsfw"Required. NSFW inputs require NSFW output.
stylestringRequired. Rendering style. Use realism (photographic) or anime; these are the values the catalog and examples use. Other text up to 64 characters is accepted and stored with the job, but no other style is supported.
adultSubjectsVerifiedbooleanRequired true for NSFW. Confirms fictional or consenting verified-adult subjects.
referenceAssetVersionIdsstring[]Required. Owned image asset versions in Picture 1 through Picture 3 order.
motion"idle" | "transition"Required.
resolution"736p"Portrait video resolution. Direct video is 736p (736 × 1280) only; omit the field or set it to 736p. 544p is no longer offered.

POST /generations/eventsread

Read committed revisions for up to 50 owned jobs; reconnect to obtain newer revisions.

Request body
FieldTypeNotes
jobIdsstring[]Required.

POST /generations/extend-to-posegenerate

Extend an owned source image to a current text- or image-guided pose; guidance is resolved internally

Request body: TargetPoseExtensionIntent (mode "extend")
FieldTypeNotes
mode"extend"Required.
requestIdstring (uuid)Required.
sourceFrameAssetVersionIdstringRequired.
duration3 | 5Required.
seedintegerRequired.
targetPoseIdstringRequired.
targetPoseRevisionstring (date-time)Required.
transitionInstructionsstringOptional user instructions for the route from the starting image to the selected text- or image-guided destination pose. Inserted during the transition; extraMotion applies after arrival.
extraMotionstring
audiostring

POST /generations/frame-to-framegenerate

Accept a first- and last-frame FL2VA job and reserve points

Request body: FrameToFrameIntent
FieldTypeNotes
requestIdstring (uuid)Required.
firstFrameAssetVersionIdstringRequired. Owned image asset version for the opening frame.
lastFrameAssetVersionIdstringRequired. Owned image asset version for the final frame. Reuse the opening image for idle motion.
duration3 | 5 | 7Required. Clip length in seconds. Output is 736 × 1280 (720p). 1 and 10 seconds are no longer offered.
seedintegerRequired.
motionPromptstringOptional motion direction between the frames.
audiostringOptional soundscape direction. Blank uses natural ambient sound and no dialogue.

POST /generations/frame-to-frame/quoteread

Quote a first- and last-frame FL2VA job

Request body: FrameToFrameIntent
FieldTypeNotes
requestIdstring (uuid)Required.
firstFrameAssetVersionIdstringRequired. Owned image asset version for the opening frame.
lastFrameAssetVersionIdstringRequired. Owned image asset version for the final frame. Reuse the opening image for idle motion.
duration3 | 5 | 7Required. Clip length in seconds. Output is 736 × 1280 (720p). 1 and 10 seconds are no longer offered.
seedintegerRequired.
motionPromptstringOptional motion direction between the frames.
audiostringOptional soundscape direction. Blank uses natural ambient sound and no dialogue.

POST /generations/imagegenerate

Generate a portrait image from a direct prompt

Request body: DirectImageGenerationIntent
FieldTypeNotes
requestIdstring (uuid)Required.
promptstringRequired. Direct image prompt, passed to the image worker as written.
seedintegerRequired.
contentRating"sfw" | "nsfw"Required. NSFW inputs require NSFW output.
stylestringRequired. Rendering style. Use realism (photographic) or anime; these are the values the catalog and examples use. Other text up to 64 characters is accepted and stored with the job, but no other style is supported.
adultSubjectsVerifiedbooleanRequired true for NSFW. Confirms fictional or consenting verified-adult subjects.
referenceAssetVersionIdsstring[]Optional ordered list of owned or eligible admin-published image versions. Use an empty list or omit the field for prompt-only generation.

POST /generations/image/quoteread

Quote a direct portrait image generation

Request body: DirectImageGenerationIntent
FieldTypeNotes
requestIdstring (uuid)Required.
promptstringRequired. Direct image prompt, passed to the image worker as written.
seedintegerRequired.
contentRating"sfw" | "nsfw"Required. NSFW inputs require NSFW output.
stylestringRequired. Rendering style. Use realism (photographic) or anime; these are the values the catalog and examples use. Other text up to 64 characters is accepted and stored with the job, but no other style is supported.
adultSubjectsVerifiedbooleanRequired true for NSFW. Confirms fictional or consenting verified-adult subjects.
referenceAssetVersionIdsstring[]Optional ordered list of owned or eligible admin-published image versions. Use an empty list or omit the field for prompt-only generation.

Help

Questions or a problem with the API: support@mytype.video. Prepayment, volume pricing and higher limits: sales@mytype.video or the business contact form. The API terms of use and the terms of service apply to API use. Price changes and deprecations are announced in the changelog and by email at least 30 days before they take effect; live service status is on the status page.

Changes. API version v1 only gains fields; existing fields keep their meaning. A change that breaks existing requests is announced the same way, at least 30 days ahead.