minimax-h3-liteMiniMax H3 Lite runs the same MiniMax H3 multimodal video model as minimax-h3, on a budget channel: 480P at $0.01 per second and 768P at $0.02 per second, any whole length from 1 to 15 seconds, so a 768P 10-second clip is $0.20 and a 480P 5-second draft is $0.05. One endpoint covers the useful modes and picks the right pipeline from what you send: a prompt alone is text-to-video; add up to 9 reference images to lock character, product or style; add up to 3 reference audio clips (with at least one image) to drive voice and lip movement; or send a first and last frame for an interpolated shot. Reference images and audio are free. Two things are deliberately left out to keep the price where it is: reference video (motion transfer) and native 2K — for those, use minimax-h3. Output is landscape 16:9 or portrait 9:16. Typical turnaround is about 1.5 minutes for a 480P 3-second clip, 2.5 minutes for 768P 5 seconds and 8 minutes for 768P 15 seconds, plus queue time. Only successful generations are charged.
768P at $0.02/s versus $0.097/s on minimax-h3 — the same MiniMax H3 weights, so the character consistency and prompt-following carry over; what you give up is 2K and reference video
Whole-second control from a 1-second test to a 15-second shot, billed per second: a 768P 10s clip is $0.20, a 480P 5s draft is $0.05
Prompt alone is text-to-video; add reference images, reference audio or a first + last frame and the request is routed to the matching pipeline — no separate model ids to learn
Up to 9 reference images and 3 reference audio clips at no extra charge — audio drives voice and lip movement when paired with at least one image
No reference video, no 2K, and 16:9 or 9:16 only — stated up front so you can pick minimax-h3 when the job needs more
Leave the references below empty for pure text-to-video — every reference input is optional.
Steer character, style or composition. Reference images are free on Lite.
MP3 / WAV, ≤15MB and 2-15s each. Drives voice and lip movement; needs at least one reference image. Free.
Generated video will appear here
Provide URLs and click Generate
MiniMax H3 Lite is a Video Generation API provided by MiniMax. MiniMax H3 Lite runs the same MiniMax H3 multimodal video model as minimax-h3, on a budget channel: 480P at $0.01 per second and 768P at $0.02 per second, any whole length from 1 to 15 seconds, so a 768P 10-second clip is $0.20 and a 480P 5-second draft is $0.05. One endpoint covers the useful modes and picks the right pipeline from what you send: a prompt alone is text-to-video; add up to 9 reference images to lock character, product or style; add up to 3 reference audio clips (with at least one image) to drive voice and lip movement; or send a first and last frame for an interpolated shot. Reference images and audio are free. Two things are deliberately left out to keep the price where it is: reference video (motion transfer) and native 2K — for those, use minimax-h3. Output is landscape 16:9 or portrait 9:16. Typical turnaround is about 1.5 minutes for a 480P 3-second clip, 2.5 minutes for 768P 5 seconds and 8 minutes for 768P 15 seconds, plus queue time. Only successful generations are charged. Through APIMODELS platform, you can access this model via a unified API with transparent pay-as-you-go pricing. Current pricing: 480P · per second: $0.01, 480P · 5s: $0.05, 480P · 10s: $0.1, 480P · 15s: $0.15, 768P · per second: $0.02, 768P · 5s: $0.1, 768P · 10s: $0.2, 768P · 15s: $0.3, Reference images (up to 9): $0, Reference audio (up to 3): $0.
Quickly generate brand promotion videos for ad campaigns and social media marketing.
Create compelling short-form video content for platforms like TikTok, Instagram, and YouTube.
Generate product feature demonstrations and tutorials to improve user conversion.
Produce course explanations, knowledge explainers, and training videos at low cost.
MiniMax H3 Lite is available through APIMODELS at: 480P · per second: $0.01, 480P · 5s: $0.05, 480P · 10s: $0.1, 480P · 15s: $0.15, 768P · per second: $0.02, 768P · 5s: $0.1, 768P · 10s: $0.2, 768P · 15s: $0.3, Reference images (up to 9): $0, Reference audio (up to 3): $0. Billing is pay-as-you-go — you only pay for what you generate.
Sign up at APIMODELS, get your API key, and call our unified API endpoint. We provide detailed API documentation with code examples in cURL, Python, and Node.js.
APIMODELS offers the same MiniMax H3 Lite model through our aggregation platform. We provide a unified API interface so you do not need separate accounts for each provider - one API key to access all models.
Same MiniMax H3 multimodal video model, different channel and price. Lite runs on a budget channel and bills per second: 480P at $0.01/s and 768P at $0.02/s, versus $0.097/s (768P) and $0.145/s (native 2K) on minimax-h3. The model-level strengths — character consistency, prompt following — carry over. What Lite leaves out is native 2K and reference video (motion transfer). Use Lite for drafts, social clips and volume; use minimax-h3 when you need 2K or want to reproduce the camera move of a reference video.
Per second, in two resolution tiers: 480P at $0.01/s (5s $0.05, 10s $0.10, 15s $0.15) and 768P at $0.02/s (5s $0.10, 10s $0.20, 15s $0.30). Any whole length from 1 to 15 seconds, billed for the seconds you ask for. Reference images (up to 9) and reference audio (up to 3 clips) cost nothing extra. No minimum top-up, no subscription, and only successful requests are charged — failures are refunded.
Use the unified video endpoint POST /api/v1/video/generations with model set to minimax-h3-lite. prompt is required; duration is any whole second from 1 to 15 (default 5), resolution is 480p or 768p (default 768p), ratio is 16:9 or 9:16 (default 16:9). Attach references as needed: images (up to 9), audio_list (up to 3, needs at least one image), or first_frame_url + last_frame_url for a first/last-frame shot — each accepts a public URL or base64. The call returns a taskId; poll GET /api/v1/video/generations?task_id=xxx or pass callback_url. Field names match minimax-h3, so switching between the two means changing model only.
No reference video (sending one returns a clear 400 at create time rather than being dropped silently); resolution tops out at 768P — no 1080P or 2K; aspect ratio is landscape 16:9 or portrait 9:16 only; first/last-frame mode needs both frames and cannot be mixed with reference images or audio; reference audio needs at least one reference image. Measured turnaround: about 1.5 minutes for 480P 3s, 2.5 minutes for 768P 5s and 8 minutes for 768P 15s, plus queue time. Result files are kept for 7 days.
On APIMODELS, MiniMax H3 Lite runs alongside 60+ models on one API key and one balance, so choosing is about fit, not lock-in. It supports 480P / 768P, Multimodal Reference, 1-15s, Up to 9 Ref Images, Up to 3 Ref Audios, First + Last Frame, From $0.01/s, and you can weigh it on price and capability against other Video Generation models, then switch by changing a single model-name string — no new account or integration. Browse every Video Generation option with live pricing at apimodels.app/models.
MiniMax H3 Lite supports: 480P / 768P, Multimodal Reference, 1-15s, Up to 9 Ref Images, Up to 3 Ref Audios, First + Last Frame, From $0.01/s. See the APIMODELS docs for full parameters and call examples.
Yes. APIMODELS exposes MiniMax H3 Lite through a single unified API and one key — no separate provider accounts, and no need to handle each provider's regional network access yourself.
We support Stripe (Visa, Mastercard, and other international cards) and Alipay. Credits are available instantly after payment.