📌 Quick answer:

• What it is: MiniMax's newest video model family, generating 5–15 second clips with native stereo audio at 480p, 768p or 1080p in six aspect ratios.
• Price on Vidgo: H3 Max costs $0.05 / $0.08 / $0.16 per second at 480p / 768p / 1080p. H3 Max Turbo is exactly half: $0.025 / $0.04 / $0.08.
• A typical clip: 5 seconds at 768p costs $0.40 on H3 Max and $0.20 on Turbo.
• Max or Turbo? Use Turbo for drafts and batch work. Use H3 Max for final shots and whenever you need reference-to-video with image, video and audio references, which Turbo does not offer.
• Try it: H3 Max and H3 Max Turbo are live on Vidgo with a browser playground and API; new accounts start with free credits.

MiniMax also makes the Hailuo video models, so teams already on Hailuo 2.3 are the most likely to evaluate H3 Max next. This guide covers what changed, what it costs, and how to call it. Specifications and prices were checked on October 13, 2026 against Vidgo's model pages and API contract. Vidgo publishes this guide; we have not run our own speed or quality benchmark.

MiniMax Video Models on Vidgo at a Glance

The quickest way to read this table: H3 Max and Turbo trade fixed clip lengths for any duration from 5 to 15 seconds, add six aspect ratios and native sound, and cover the 480p–1080p range. The original H3 stays relevant only when you need 2K output.

---

MiniMax H3 Max API Pricing

Both models bill per second of requested duration, with the rate set by resolution. One Vidgo credit is $0.005.

Image-to-video costs the same as text-to-video, and adding an end frame has no extra charge. Reference-to-video (H3 Max only) costs the output rate above plus reference usage: the first 4,096 reference tokens per task are free, then 4 credits ($0.02) per 1,000 tokens, prorated. Budget a little headroom for reference jobs, because that usage is metered by the server after submission.

Example budget: 200 social clips a month at 5 seconds and 768p cost $80 on H3 Max or $40 on Turbo. Drafting all 200 on Turbo at 480p ($25) and re-rendering the best 40 on H3 Max at 1080p ($32) totals $57.

Sources: H3 Max pricing and H3 Max Turbo pricing on Vidgo.

---

H3 Max vs H3 Max Turbo

Turbo is a distilled version of H3 Max. According to its model page, it generates about twice as fast and costs half as much at every resolution.

Two details matter in practice. First, Turbo's model page says its 1080p tier is upsampled from 768p, so 768p is Turbo's best value; if you need true 1080p detail, render the final take on H3 Max. Second, only H3 Max accepts reference images, clips and audio, so any workflow that must keep a character or product consistent across shots needs H3 Max.

H3 Max vs the Original H3 and Hailuo 2.3

Original H3. Choose it only if you need 2K output: Vidgo's H3 route is fixed at 2K for $0.105 per second. For anything delivered at 1080p or below, H3 Max at 768p is cheaper per second ($0.08) and Turbo cheaper still.

Hailuo 2.3. Hailuo 2.3 is priced per clip with fixed lengths: $0.175 for 6 seconds at 768p, $0.35 for 10 seconds, and $0.30 for 6 seconds at 1080p. At 768p that is about $0.029 per second, cheaper than H3 Max Turbo's $0.04. Stay on Hailuo 2.3 if 6- or 10-second silent clips are all you need. Move to H3 Max or Turbo when you need native audio, other aspect ratios, durations other than 6 or 10 seconds, or reference-guided generation.

---

The Three Generation Modes

Text to video

Send a prompt (up to 50,000 characters), a duration of 5–15 seconds, a resolution, and one of six aspect ratios: 21:9, 16:9, 4:3, 1:1, 3:4 or 9:16 (default 16:9). Describe sound as well as visuals; the model generates matching stereo audio in the same pass.

Image to video

Send one or two images: the first is the opening frame, the optional second is the closing frame. H3 Max's page also lists lip-synced dialogue for speaking characters in close-ups and medium shots. This mode has no aspect-ratio field, so prepare source images in the framing you want to deliver.

Reference to video (H3 Max only)

Combine 1–9 reference images with up to 3 reference videos and up to 3 reference audio files, with no more than 12 files in total. Reference clips must each run 2–15 seconds, with a combined limit of 15 seconds per media type. Aspect ratio defaults to adaptive, or you can set one of the six fixed ratios. Use this mode to keep a character, product or motion rhythm consistent across shots.

How to Call the MiniMax H3 Max API

Every Vidgo media model uses the same asynchronous flow: submit a task, then poll its status or receive a callback.

curl -X POST https://api.vidgo.ai/api/generate/submit \
  -H "Authorization: Bearer $VIDGO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "minimax/h3-max/text-to-video",
    "callback_url": "https://your-domain.com/api/webhook",
    "input": {
      "prompt": "Slow dolly through a rainy neon night market, vendors calling out, sizzling woks, rain on canvas awnings",
      "duration": 5,
      "resolution": "768p",
      "aspect_ratio": "16:9"
    }
  }'

curl https://api.vidgo.ai/api/generate/status/YOUR_TASK_ID \
  -H "Authorization: Bearer $VIDGO_API_KEY"

To switch models, change only the model string:

Image-to-video takes image_urls; reference-to-video takes reference_image_urls, reference_video_urls and reference_audio_urls. Media must be public HTTP(S) URLs. Full field lists are in the text-to-video, image-to-video, reference-to-video and Turbo docs.

Prompting Tips

• Write in layers: subject and setting, then the action, then camera movement and shot size, then lighting and sound.
• One main event per clip. A 5–8 second shot works best around a single coherent action.
• Name the sounds you want, such as footsteps on wood or rain on a car window, to improve audio-visual alignment.
• Match composition to the ratio. Emphasize depth for 21:9 and vertical framing for 9:16.
• Iterate cheap, finish sharp. Compare takes on Turbo at 768p, then render the chosen shot on H3 Max at 1080p.

Which MiniMax Model Should You Use?

What are you producing?
 │
 ├── Drafts, A/B variants, high-volume social clips ──► 【H3 Max Turbo, 768p】
 ├── Final 1080p delivery ────────────────────────────► 【H3 Max, 1080p】
 ├── Consistent character or product across shots ────► 【H3 Max reference-to-video】
 ├── Talking character from a still image ────────────► 【H3 Max image-to-video】
 ├── Native 2K output ────────────────────────────────► 【MiniMax H3 (original)】
 └── Fixed 6 s / 10 s silent clips at lowest cost ────► 【Hailuo 2.3】

Where to Access MiniMax H3 Max

Vidgo offers all five H3 Max and Turbo endpoints with a browser playground, shows per-second pricing before you submit, and uses the same submit endpoint as Veo, Kling, Seedance and Nano Banana.

FAQ

What is MiniMax H3 Max?

MiniMax H3 Max is a video generation model that creates 5–15 second clips with native stereo audio from text, images, or a mix of image, video and audio references, at 480p, 768p or 1080p.

How much does the MiniMax H3 Max API cost?

On Vidgo, H3 Max costs $0.05, $0.08 or $0.16 per second at 480p, 768p or 1080p. H3 Max Turbo costs half: $0.025, $0.04 or $0.08 per second. Reference-to-video adds metered reference usage beyond the first 4,096 tokens.

What is the difference between H3 Max and H3 Max Turbo?

Turbo is a distilled version that runs about twice as fast at half the price. It supports text-to-video and image-to-video but not reference-to-video, and its 1080p tier is upsampled from 768p.

How long can MiniMax H3 Max videos be?

Any whole number of seconds from 5 to 15. Billing follows the requested duration.

Does MiniMax H3 Max generate audio?

Yes. Both H3 Max and Turbo generate synchronized stereo audio with the video. Describe the sounds you want in the prompt.

Is MiniMax H3 Max the same as Hailuo?

MiniMax makes both. Hailuo 2.3 and Hailuo 02 are earlier MiniMax video models with fixed clip lengths and no listed native audio on Vidgo; H3 Max is the newer family with flexible durations, six aspect ratios and native sound.

---

👉 Try MiniMax H3 Max on Vidgo or start cheaper with H3 Max Turbo. For other video models, see Best AI Video Generators in 2026.