Drop files here, paste files, or add a URL. 1–2 images, ordered as first frame then optional last frame. JPEG, PNG or WebP; up to 30 MB each.
Drop files here, paste files, or add a URL. 1–9 reference images. JPEG, PNG or WebP; up to 30 MB each. Supported ratios: 1:1, 4:3/3:4, 16:9/9:16 and 5:2/2:5. At most 12 reference assets combined.
Drop files here, paste files, or add a URL. Up to 3 MP4/MOV videos, 50 MB each; requires a reference image. Each must be 2–15s, with at most 15s of video combined. Constant 24fps, 16:9 or 9:16, square pixels and no rotation metadata. After clipping to output duration, the frame count must be 48, 120, 240 or 360; verified by the server.
Drop files here, paste files, or add a URL. Up to 3 MP3/WAV audio files, 15 MB each; requires a reference image. Each must be 2–15s and no longer than the output, with at most 15s of audio combined. Only 32kHz with a sample count divisible by 800 is supported; verified by the server.
Example Output
This is sample data. Generate a new result to see real output.
Transparent pricing with no hidden fees. Pay as you go.
h3-maxh3-maxh3-max| Model | Spec | PoYo price | Official price | You save |
|---|---|---|---|---|
| h3-max | 480p | $0.050/sec 10 credits /sec | - | - |
768p | $0.080/sec 16 credits /sec | - | - | |
1080p | $0.160/sec 32 credits /sec | - | - |
* Actual fees are based on the final output.
High-throughput video generation, superior prompt fidelity, and native synchronized stereo audio
Access MiniMax H3 Max API instantly on PoYo—no waitlist required. Experience faster-than-real-time generation, exceptional prompt adherence, and native synchronized stereo audio in a single pass.
Create 5–15 second cinematic videos at 480p, 768p, or 1080p with flexible text-to-video, first/last frame guidance, and multimodal reference workflows.
Production-ready features for high-throughput video generation, superior prompt fidelity, native audio workflows, and precision frame control.
01
The Last Signal follows a station attendant lifting his lantern as the final train leaves a rainy platform. This 5-second, 768p request used the same prompt as the Turbo example. The recorded test took 21 seconds from task creation to completion and cost 80 credits; timings vary with queue and service conditions.
02
Lost on the Platform follows a brass pocket watch as it drops, rolls through a puddle and is caught at the platform edge. The 8-second, 21:9 example uses a low tracking camera to stage an ordered action with water, weight and a clear resolution. The 1080p tier returned a 2200 × 942 widescreen file.
03
Someone Answers begins with a single reference image of the lighthouse keeper. The 8-second, 1080p close-up prompts the line “I can see your light.” followed by a distant foghorn, combining restrained facial performance with native dialogue and environmental sound.
04
Homeward combines two reference images with a 2-second, 24fps video and 32kHz audio excerpt from the lighthouse story. The keeper steps outside and watches a boat approach through the beam. This 10-second, 768p example demonstrates multimodal guidance separately from the default first/last-frame scene. Reference fees are calculated by the server in addition to base output credits.
Generate engaging vertical 9:16 videos with synchronized speech and ambient sound for TikTok, Reels, and Shorts at high velocity.
Rapidly storyboard ideas, animate character concept sketches with first/last frames, and iterate on visual styles in seconds.
Produce breaking news visuals, factual video explainers, and editorial video clips with synchronized sound in real time.
Create high-converting promotional ads, localized variations, and multi-ratio campaigns at scale without expensive production shoots.
Prototype in-game cinematics, creature animations, environment reveals, and promotional cutscenes with controlled camera movements.
Turn instructional scripts into audiovisual demonstrations with talking avatars, clear voice alignment, and visual explanations.
Compare MiniMax H3 Max, MiniMax H3 Max Turbo, and the parent MiniMax H3 (Hailuo 03) model to choose the right engine for generation speed, budget, output resolution, and creative reference control.
| Feature | MiniMax H3 Max | MiniMax H3 Max Turbo | MiniMax H3 (Hailuo 03) |
|---|---|---|---|
| Primary Positioning | High-throughput production & rapid iteration | High-throughput production & cost efficiency | High-resolution 2K/4K mastering & multimodal references |
| Generation Speed | 21s in this test; latency varies | 17s in this test; latency varies | Standard production speed |
| Prompt Adherence | Superior (fal-optimized post-training) | Strong prompt adherence | High fidelity |
| Supported Resolutions | 480p, 768p, 1080p | 480p, 768p, 1080p | 480p, 768p, native 2K |
| Audio Generation | Native stereo audio with lip-sync | Native stereo audio with lip-sync | Native stereo audio with lip-sync |
| Frame Control | First and optional last frame | First and optional last frame | First and optional last frame |
| Pricing on PoYo | 10–32 credits/sec (by resolution) | 5–16 credits/sec (50% of H3 Max output pricing) | Based on duration & resolution |
Choose MiniMax H3 Max for rapid iteration, high-volume production, and multimodal references. Choose MiniMax H3 Max Turbo for faster generation and lower output costs. Choose MiniMax H3 for native 2K mastering and multimodal reference conditioning.
Step 1: Register & Create API Key
Sign up on PoYo and generate your API key instantly in the dashboard with no waitlist.
Create API Key →
Step 2: Top Up Credits
Add credits to your balance with flexible pay-as-you-go pricing and zero recurring subscriptions.
Add Credits →
Step 3: Integrate the API
Submit asynchronous video tasks via REST API with model h3-max and fetch completed MP4 videos.
View Documentation →
Start generating videos immediately with straightforward REST endpoints and instant API key provisioning, bypassing approval queues.
No recurring subscriptions or minimum commitments. Pay strictly for generated seconds and actual token consumption with competitive credit rates.
Engineered on high-availability cluster infrastructure with 99.9% uptime, reliable task queuing, and fast content delivery network.
Interactive playground, verified request schemas, comprehensive SDK examples, and responsive technical support for rapid production rollout.
