VidMage

VidMage is an AI video and image creation platform for creators. Turn your ideas and images into videos and fresh visuals.

Open the creator tools
Explore APIs
AI Video APIsAI Image APIsAI Audio APIsAI 3D APIsAI Face Swap APIsAI Effects APIs
Build with VidMage
QuickstartAPI keysVidMage MCPError reference
Your account
Developer consoleAbout VidMagePlans and creditsAPI billingContact support
© 2026 VidMage. All rights reserved.
Privacy PolicyTerms of ServiceReport Abuse
Skip to content
VidMage/Developers
Overview
APIs
All APIsAI Video APIsAI Image APIsAI Audio APIsAI 3D APIsAI Face Swap APIsAI Effects APIs
DocumentationVidMage MCPOpen console
Developers/MiniMax H3 API

MiniMax H3 API

Build video generation workflows with MiniMax H3 using text, starting and ending frames, or image, video, and audio references. Configure clip duration and resolution in the request to match your application.

Get API keyView API docs
Abstract illustration for MiniMax H3 API
PlaygroundAPIPricingGuideFAQs

MiniMax H3 API Playground

Use your existing API key and subscription. Submitting a generation uses credits; uploading and preparing a request does not start a generation.

API access is available to subscribers only. Your subscription credits are shared across the website and the API.

View plans

Parameter

10 parameters

Text prompt describing the video to generate (max 20480 chars).

Optional input image URL. When provided, the task runs in image-to-video mode. In that mode, output ratio is inferred from the input image.(Optional)

Optional multimodal reference images. Public URLs only; 30 MB per file.(Optional)

maxItems 9

Optional multimodal reference videos. Each video is 50 MB max.(Optional)

maxItems 3

Optional multimodal reference audio. Each audio file is 15 MB max.(Optional)

maxItems 3

Optional public URL for the final frame in image-to-video mode.(Optional)

Video duration in seconds.(Optional)

Default: 5

Output aspect ratio for modes that expose a ratio parameter. Ignored when imageUrl selects a mode whose ratio is inferred from the input image.(Optional)

Default: 16:9

Output resolution tier.(Optional)

Default: 2k

Add an AIGC watermark to the generated video.(Optional)

Default: false

Request code

#!/usr/bin/env bash
set -euo pipefail
# Set this once per intended task; preserve it and the body for transport retries.
: "${VIDMAGE_IDEMPOTENCY_KEY:?Set a unique key for this task}"

SUBMIT_RESPONSE="$(curl --fail-with-body --silent --show-error -X POST "https://vidmage.ai/api/v1/minimax-h3-ai-video-generator/submit" \
  -H "Authorization: Bearer ${VIDMAGE_API_KEY}" \
  -H "Idempotency-Key: ${VIDMAGE_IDEMPOTENCY_KEY}" \
  -H "Content-Type: application/json" \
  --data-raw '{
  "prompt": "A native 2K product film where the reference image controls the object, the reference clip controls the camera move, and the final frame lands on a clean hero composition",
  "duration": "5",
  "aspectRatio": "16:9",
  "resolution": "2k",
  "aigcWatermark": false
}')"
TASK_ID="$(printf '%s' "$SUBMIT_RESPONSE" | jq -er '.["taskId"]')"

while true; do
  QUERY_RESPONSE="$(curl --fail-with-body --silent --show-error -X POST "https://vidmage.ai/api/v1/minimax-h3-ai-video-generator/query" \
    -H "Authorization: Bearer ${VIDMAGE_API_KEY}" \
    -H "Content-Type: application/json" \
    --data-raw "{\"taskId\":\"${TASK_ID}\"}")"
  STATUS="$(printf '%s' "$QUERY_RESPONSE" | jq -r '(.status // .data.status // "") | ascii_downcase')"
  case "$STATUS" in
    success|succeeded|completed)
      RESULT="$(printf '%s' "$QUERY_RESPONSE" | jq -r '(.result // .["videoUrl"] // .data["videoUrl"] // empty)')"
      if [ -z "$RESULT" ]; then
        printf '%s' "Result missing; preserve task $TASK_ID and query the same task again. Do not resubmit." >&2
        exit 2
      fi
      printf '%s\n' "$RESULT"
      break
      ;;
    needs_input)
      printf '%s' "Face selection required; preserve task $TASK_ID" >&2
      exit 2
      ;;
    failed|error)
      printf '%s' "$QUERY_RESPONSE" | jq -r '(.message // .error // .data.error // "Task failed")' >&2
      exit 1
      ;;
  esac
  sleep 5
done

Response data

Submit the task to see the API response here.
PlaygroundCapabilities

API documentation

Submit a request, track the task, and retrieve your result.

Required inputs
prompt
Output
Video·videoUrl
Request setup & limitsHeaders, input options and media limits

Request headers

Authorization
Bearer YOUR_API_KEY

Keep your API key on your server.

Content-Type
application/json
Idempotency-Key
YOUR_UNIQUE_KEY

Use a new key per task. Reuse it only when retrying the same submission.

Input options & limits

Request controls follow each task's parameter rules; they do not guarantee output properties.

MiniMax H3 AI Video Generator

Modes: text-to-video, image-to-video, multimodal-video

Additional inputs; requirements vary by mode

prompt imageUrl imageUrls videoUrls audioUrls lastFrameUrl

Media limits by mode
  • Image To VideoImage: jpg, jpeg, png, webp
  • Multimodal VideoImage: jpg, jpeg, png, webp; up to 30 MB per file; up to 9 filesVideo: mp4, mov; up to 50 MB per file; up to 3 filesAudio: mp3, wav; up to 15 MB per file; up to 3 files
Request controls
  • duration: 11 accepted values; see parameter rules
  • aspectRatio: "16:9", "9:16", "1:1", "4:3", "3:4", "21:9", "adaptive"
  • resolution: "768p", "2k"
  • aigcWatermark: default false

Input retention and output-link lifetime depend on the API contract. Confirm API-specific retention terms before making promises to your users. Upload guide ↗

MiniMax H3 AI Video Generator

POST /api/v1/minimax-h3-ai-video-generator/submit

Full API documentation ↗

prompt is required.

Required inputs

promptstring
Text prompt describing the video to generate (max 20480 chars).

Edit the sample inputs for your own task before submitting.

MiniMax H3 AI Video Generator request
# Set VIDMAGE_IDEMPOTENCY_KEY to a unique value for this task; preserve it for transport retries.
curl -X POST "https://vidmage.ai/api/v1/minimax-h3-ai-video-generator/submit" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Idempotency-Key: ${VIDMAGE_IDEMPOTENCY_KEY}" \
  -H "Content-Type: application/json" \
  -d '{"prompt":"A native 2K product film where the reference image controls the object, the reference clip controls the camera move, and the final frame lands on a clean hero composition","duration":"5","aspectRatio":"16:9","resolution":"2k","aigcWatermark":false}'
# -> { "success": true, "taskId": "...", "creditsConsumed": ... }
Parameters and input rules10 fields
FieldTypeRequiredDefaultMeaning and limits
promptstringYesNot specifiedText prompt describing the video to generate (max 20480 chars). Minimum length: 1 · Maximum length: 20480
imageUrlstringNoNot specifiedOptional input image URL. When provided, the task runs in image-to-video mode. In that mode, output ratio is inferred from the input image. No additional field constraint listed.
imageUrlsstring[]NoNot specifiedOptional multimodal reference images. Public URLs only; 30 MB per file. Maximum items: 9
videoUrlsstring[]NoNot specifiedOptional multimodal reference videos. Each video is 50 MB max. Maximum items: 3
audioUrlsstring[]NoNot specifiedOptional multimodal reference audio. Each audio file is 15 MB max. Maximum items: 3
lastFrameUrlstringNoNot specifiedOptional public URL for the final frame in image-to-video mode. No additional field constraint listed.
durationstringNo"5"Video duration in seconds. Values: "5", "6", "7", "8", "9", "10", "11", "12", "13", "14", "15" · Default: "5"
aspectRatiostringNo"16:9"Output aspect ratio for modes that expose a ratio parameter. Ignored when imageUrl selects a mode whose ratio is inferred from the input image. Values: "16:9", "9:16", "1:1", "4:3", "3:4", "21:9", "adaptive" · Default: "16:9"
resolutionstringNo"2k"Output resolution tier. Values: "768p", "2k" · Default: "2k"
aigcWatermarkbooleanNofalseAdd an AIGC watermark to the generated video. Default: false
Mode-specific fields & media limits
Text To Video rules
ControlMode-specific rule
promptRequired; minimum 1 characters; maximum 20480 characters
durationValues: "5", "6", "7", "8", "9", "10", "11", "12", "13", "14", "15"; mode default: "5"
aspectRatioValues: "16:9", "9:16", "1:1", "4:3", "3:4", "21:9"; mode default: "16:9"
resolutionValues: "768p", "2k"; mode default: "2k"
aigcWatermarkDefault: false
Image To Video rules
ControlMode-specific rule
promptRequired; minimum 1 characters; maximum 20480 characters
durationValues: "5", "6", "7", "8", "9", "10", "11", "12", "13", "14", "15"; mode default: "5"
resolutionValues: "768p", "2k"; mode default: "2k"
imageUrlImage input
lastFrameUrlOptional final-frame image input
aigcWatermarkDefault: false
Multimodal Video rules
ControlMode-specific rule
promptRequired; minimum 1 characters; maximum 20480 characters
durationValues: "5", "6", "7", "8", "9", "10", "11", "12", "13", "14", "15"; mode default: "5"
aspectRatioValues: "adaptive", "16:9", "9:16", "1:1", "4:3", "3:4", "21:9"; mode default: "adaptive"
resolutionValues: "768p", "2k"; mode default: "2k"
imageUrlsImage input; up to 9 items
videoUrlsVideo references; up to 3 items
audioUrlsAudio references; up to 3 items
aigcWatermarkDefault: false

Retrieve the result

Save taskId from the accepted submission and send it to this endpoint:

POST /api/v1/minimax-h3-ai-video-generator/query

Use the same Bearer API key. On completion, read the videoUrl result field.

Task states and recovery ↗
MiniMax H3 AI Video Generator query
curl -X POST "https://vidmage.ai/api/v1/minimax-h3-ai-video-generator/query" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "taskId": "TASK_ID_FROM_SUBMIT" }'
# -> { "success": true, "data": { "status": "...", "videoUrl": "https://..." } }

Pricing

See what one image, video, or generation costs in credits. Your API and website activity share the same credit balance.

API credit rates and calculated example request costs
API / workflowExample requestPrice (credits)
MiniMax H3 AI Video Generatorvideo API5-second video: 125 creditsresolution: 2k
How this is calculated
  • 768p: 15 credits / second
  • 2k: 25 credits / second

Base charge = duration × the selected resolution rate.

Each reference image beyond 5 adds 10 credits.

Estimate your own request ↗
15–25 credits / secondBase rate varies by settings or workflow
Billed in credits. Shared with the website.Active subscription required. Unit rates and examples follow current API credit rules.Billing details⌄

Examples show the charge for the inputs listed above. Resolution, duration, audio, output count and reference media can change the total. Minimum charges and rounding follow the selected API.

Use the Playground to estimate your request before submitting it. Estimates do not start a generation. Subscription and credit-pack prices are listed on the plans page.

Your task’s recorded usage is the source of truth for the final charge.

View plans ↗Billing guide ↗

About MiniMax H3 API

Character images, a location reference, and a performance clip can all contribute to a short drama scene. VidMage's MiniMax H3 API accepts text, starting and ending frames, and image, video, or audio references for video generation. Duration, resolution, and watermark settings control the requested output. Audio inputs provide reference material, and your editing workflow assembles the selected scenes. To explore this model's browser workflow, open VidMage's MiniMax H3 AI Video Generator.

MiniMax H3 API capabilities

Define starting and ending frames
Supply a starting image and optional ending image, with a prompt explaining the motion between them.
Combine reference media
Guide a scene with up to nine image references, three video references, and three audio references.
Set clip output options
Choose a listed five- to fifteen-second duration, a 768p or 2k resolution, and the watermark setting.

What you can build

Reference-based short drama scenes

Character art explains who is in the scene; a location image shows where; a performance clip suggests the movement. Bring those references into one clearly described request to develop a scene for a short drama episode.

Commercial concept films

An agency can test a product shot using the same materials it would put in a creative brief: product stills, a camera reference, and an audio mood reference. The generated footage gives the team something concrete to discuss before the final edit.

Storyboard transitions

Connect two planned views, such as a closed storefront and its door opening. Supply the opening and ending images with a description of the action between them. The transition clip can help establish timing in an animatic.

How to use MiniMax H3 API

  1. Get an API key

    Create a key in the Developer Console and store it on your server. Use it in the Authorization: Bearer header.

  2. Prepare and submit your inputs

    Set up your inputs in the Playground, then copy the matching API request. Send it from your server with a unique Idempotency-Key.

  3. Track the task

    Save the returned taskId and query the same operation until it completes. Keep that identifier if your app stops waiting.

  4. Retrieve the result

    Read videoUrl from the completed task. Preview the result in your app and save a copy to your own storage for later use.

Input tips
Give each reference a job
Use imageUrl for a starting frame and the plural fields for reference packs. Explain which source guides the subject, setting, movement, or audio in your prompt.
Keep audio input separate
audioUrls supplies reference material; this capability does not expose a separate sound switch. Use the exact resolution values 768p or 2k when configuring output.

Input requirementsUpload local files

Task fields and results

Keep the task identifier with its original request and Idempotency-Key. Query the same capability until it completes, then read the documented result field.

Task identifier
taskId
Query route
POST /api/v1/minimax-h3-ai-video-generator/query
Completed result
videoUrl
Task states and response structure ↗
Errors, retries and limits

Retry transport failures with the same idempotency key only when the original submission may not have reached the server. For a confirmed task, keep querying the original task instead of submitting a duplicate.

Read error recovery guidance ↗

FAQs

Is this a self-hosted MiniMax H3 endpoint?

No. This page describes the hosted VidMage interface. Native-provider examples and local inference commands use different request contracts.

Are failed or timed-out requests charged?

A client timeout is not a confirmed task failure. Confirmed failures are automatically refunded only when the refund outcome is certain. Unknown charges or refunds remain pending reconciliation.

Failure and refund handling ↗
How long should my app wait for a result?

The checked contract does not establish a fixed completion time. Choose a local wait budget for your app; it is not an API completion deadline.

Polling and wait budgets ↗
What are the rate and concurrency limits?

Request-rate limits control how often you can call the API; concurrency limits control simultaneous work. The documentation ties both to the stable API key but does not publish numeric limits for this operation. Confirm account limits before planning parallel jobs.

Rate-limit recovery ↗
Can I receive results through a webhook?

The checked request contract documents status queries and does not list a webhook or callback field for these routes. Check the current authenticated OpenAPI contract before making a callback part of your integration.

Query task results ↗Check the current contract ↗

Explore more APIs on VidMage

  • HappyHorse API↗
  • Sora 2 API↗
  • Runway API↗
  • Seedance API↗
  • Seedance 2.5 API↗
  • PixVerse API↗
  • Hailuo API↗
  • Midjourney Video API↗