Low-cost prompt experiments, draft composition, and batch creative testing; $0.003 for 1 second.
Cheap
MiniMax H3 API
Create text-to-video and image-to-video jobs through a small asynchronous REST API. Fast 768p starts at $0.003 per output second, with no monthly API fee and no required client SDK.
curl https://api.vgenv.com/v2/videos \
-H "Authorization: Bearer $VGENV_API_KEY" \
-H "Idempotency-Key: h3-demo-001" \
-H "Content-Type: application/json" \
-d '{
"model": "minimax/h3",
"tier": "fast",
"prompt": "A cinematic paper city waking at sunrise",
"duration": 5,
"resolution": "768p",
"aspect_ratio": "9:16"
}'Low-cost 768p pricing for prototypes and production
Fast 768p is the lowest displayed price tier at $0.003 per generated second: $0.003 for 1 second, $0.015 for 5 seconds, and $0.045 for 15 seconds. Prices below are vgenv global-market display rates in USD; the price snapshot returned by an accepted Video request is the source of truth for that job.
A practical production tier for regular generation workloads.
OpenVDN 8-step output for a stronger quality/speed balance.
Higher-quality output for selected production shots and finished creative assets.
See the full MiniMax H3 API pricing guide and 5–15 second cost examples
A predictable asynchronous API workflow
Every accepted request returns a Video ID, quote snapshot, and queue state. Poll the same resource until it succeeds, fails, or is cancelled.
Create
Submit the model, tier, prompt, duration, resolution, aspect ratio, and optional image inputs with an Idempotency-Key.
Poll
Read queued, running, progress, queue position, and terminal state without holding one long HTTP request open.
Store
On success, copy the result from its signed download URL before the returned expiration time.
Text, frames, and ordered reference images
The public contract validates input combinations before routing them to a published execution profile.
| Workflow | Request input | Best for |
|---|---|---|
| Text-to-video | Prompt without images | Scenes, motion, camera, and audio described in text |
| First/last-frame video | frame_images | Controlling an opening frame and optional ending frame |
| Reference-image video | reference_images | Preserving subject, product, scene, or style guidance |
Go from comparison to working code
Each guide answers one specific MiniMax H3 API search intent and links to runnable examples and the full contract.
Compare vgenv MiniMax H3 API rates by tier and resolution, calculate 5–15 second video costs, and understand quote snapshots and failure billing.
Read guidePYTHON TUTORIAL · 9 min readMiniMax H3 API with Python: create, poll, and downloadCall the vgenv MiniMax H3 video API from Python with no SDK: create an idempotent task, poll status, handle 429 responses, and read the result URL.
Read guideNODE.JS TUTORIAL · 8 min readMiniMax H3 API with Node.js and TypeScriptUse Node.js fetch with the vgenv MiniMax H3 API to create a video, poll an asynchronous task, respect Retry-After, and retrieve the signed output URL.
Read guideIMAGE-TO-VIDEO · 8 min readMiniMax H3 image-to-video API: frames and reference imagesBuild MiniMax H3 image-to-video requests with first and last frames or ordered reference images, including input limits, JSON examples, and prompt guidance.
Read guideAPI COMPARISON · 8 min readvgenv MiniMax H3 API vs the first-party MiniMax APIUnderstand how vgenv's independent MiniMax H3 API differs from MiniMax's first-party API in endpoints, model IDs, pricing, resolutions, task resources, and integration choice.
Read guideQuestions developers ask before integrating
How much does the MiniMax H3 API cost?
vgenv Fast 768p generation costs $0.003 per output second: $0.003 for one second, $0.015 for five seconds, and $0.045 for fifteen seconds. Standard 768p is $0.0045 per second and Quality 768p is $0.012 per second. The accepted Video response contains the final quote snapshot.
Is this the official MiniMax API?
No. vgenv is an independent API service. vgenv keys, endpoints, billing, model identifiers, and response fields are separate from MiniMax's first-party platform.
Does the API support image-to-video?
Yes. Use frame_images for first and optional last frames, or reference_images for up to nine ordered reference images. The two input patterns cannot be combined in one request.
Do failed generations cost money?
The quote is reserved when a job is accepted and settled only after success. Platform failures and queued cancellations release the reserved Project balance.
Do I need a Node.js or Python SDK?
No. The API uses standard Bearer authentication, HTTP, and JSON. Runnable curl, Node.js, and Python examples are available on GitHub.
Send your first low-cost request
Create a server-side API key, add Project balance, and start with a one-second or five-second Fast 768p test.
vgenv pricing reviewed August 25, 2026. vgenv is an independent service and is not affiliated with or endorsed by MiniMax. MiniMax and related product names belong to their respective owners.