Pizzeria holiday catering clip with native audio: H3 Max prices

A pizzeria can promote holiday office catering with a MiniMax H3 Max clip, which has native stereo audio, and an 8-second clip at 768p costs $0.80 on Sume. The model accepts 5 to 15 seconds and offers 480p, 768p and 1080p, so you can pick the resolution to match the placement.

Catering orders for December parties come from office managers, who buy on reliability. The clip should show food that looks hot and fresh, plus one line of text on how to order. This post covers the model's limits, what the price is at each resolution, and how to get the order line onto the video.

What H3 Max offers

From the Video Generation docs (read 2026-10-10): MiniMax H3 Max (minimax-h3-max) is the faster 768p variant. It does text-to-video, first and last frame image-to-video, and reference-to-video at 480p, 768p and 1080p for 5 to 15 seconds, with native stereo audio. 1080p is a latent refinement from native 768p. It accepts image, video and audio references as inputs.

Native audio means a sizzle, a knife on a board, a box opening. You describe the sound in the prompt next to the picture. Do not rely on it for speech or a jingle that must be exact.

• Duration: 5 to 15 seconds.
• Resolutions: 480p, 768p, 1080p (1080p refined from 768p).
• Modes: text-to-video, first/last frame, reference-to-video.

Price per second

Sume bills the provider list times 1.25. The H3 Max list rates are $0.05 per second at 480p, $0.08 at 768p and $0.16 at 1080p, which makes Sume's prices $0.0625, $0.10 and $0.20 per second (read 2026-10-10). An 8-second clip is $0.50, $0.80 or $1.60.

Ten seconds, a good length for a catering teaser, is $0.63, $1.00 or $2.00.

From your own photo

If you have a photograph of a real tray, use first-frame image-to-video so the food in the clip is your food. A text-only prompt invents a pizza that may not look like yours. The earlier post on first frame, end frame and limits explains the choice. Describe the motion, such as a lid lifting and steam rising, and the sound, such as a soft box thud and a sizzle.

Add the order line with video captions using cues, since the generated audio has no speech for the transcriber. It is $0.20 per clip, so a 10-second 768p clip with a caption is $1.20.

Choosing a resolution by placement

For a phone story, 768p is the sweet spot: it is the model's native size and avoids the refinement step that 1080p adds. Use 1080p only when the clip will be projected, for example on a counter display, or when a platform asks for it. The 480p tier is a good proof pass, at $0.50 for 8 seconds, to check that the food looks right before you pay for the final.

Run two or three variations if you can. Change one thing at a time, such as the lid opening versus a slice being lifted, and compare. Since each 8-second 768p try is $0.80, three tries are $2.40, which is a small test budget. Keep the winner and move on to the next product.

Keep the offer accurate

A catering ad makes promises about quantity, timing and price. Do not let a generated clip imply a tray size or topping count you do not sell. Say the minimum order and lead time in the caption. Test the clip at 480p first to check the camera move and the sound, then render the final at 768p. For a larger set of items, the cost-planning pattern in the ten SKU pilot post carries over.