MiniMax Hailuo 2.3 generates video with notably smooth, natural motion — characters move with weight, environments feel coherent across frames. Standard (text-to-video) and Fast (image-to-video) modes, 768p / 1080p, 6 or 10 second clips. Default settings already produce strong results.
Hailuo 2.3 is MiniMax's general-purpose video model. Where it consistently outperforms: smooth character motion (no rubber-banding limbs, weight transfer reads naturally), environmental coherence across the clip (no morphing background), and reliable output from short prompts. It's the model to pick when you want a clip that just works on the first try.
Two modes: Standard (`minimax/hailuo-2.3`) for text-to-video, Fast (`minimax/hailuo-2.3-fast`) for image-to-video from a starting frame. 768p and 1080p resolutions across 16:9, 9:16, 1:1 aspects. 6s or 10s durations (10s available at 768p only). Pricing scales with resolution and duration — 4–20 credits per clip.
2
Modes (Standard · Fast)
6 · 10s
Clip durations
768p · 1080p
Resolution tiers
Characters move with weight, environments stay coherent frame-to-frame, no rubber-banding limbs or morphing backgrounds. Hailuo handles the everyday human motion (walking, gesturing, turning) more reliably than most video models.
Standard mode generates from text alone — describe the scene and Hailuo composes the shot. Fast mode is image-to-video: upload a starting frame and Hailuo writes motion onto it. Same prompt language for both; same quality bar.
768p (default) is the right choice for social, internal review, and most web use. 1080p is for hero shots and final deliverables. 10-second clips only available at 768p — 1080p caps at 6s.
Even a 15-word prompt produces a watchable clip with intentional camera work and natural pacing. Hailuo fills in cinematography decisions without you having to over-prompt the lighting, color, or camera direction.
Each example shows the exact prompt that produced the result. Copy any prompt with one click.
Standard · 1080p · 16:9 · 6s
Wide cinematic shot of a lone fishing boat drifting across a glassy lake at golden hour, mist clinging to distant pine-covered shores, the camera does a very slow forward dolly, warm amber color grade, quiet atmospheric mood
Lead with one dominant camera move ("slow forward dolly", "static locked-off", "handheld follow") and let Hailuo handle the rest. The model picks pacing well on its own — overspecifying camera language can fight its defaults.
Standard · 768p · 16:9 · 6s
Medium shot of a parkour runner sprinting toward a rooftop edge, jacket flapping in the wind, the runner leaps the gap between buildings in mid-frame, lands in a roll and keeps running offscreen, dynamic handheld camera, city lights blurred behind
For character action, narrate the beat sequentially: setup → leap → landing. Hailuo executes the motion arc cleanly — the runner's weight on the landing is one of the things this model does better than most.
Standard · 1080p · 1:1 · 6s
Studio product shot of a matte-black ceramic coffee mug on a polished walnut surface, steam curling upward, the camera does a slow 90-degree rotational pan around the mug, soft top-down rim lighting against a dark gradient backdrop, advertising quality
Square 1:1 with a single explicit camera move (rotational pan, dolly in, push back) is the canonical product-spot setup. Hailuo at 1080p handles material textures (matte ceramic, polished walnut) very cleanly.
Standard · 1080p · 16:9 · 6s
Wide cinematic landscape of a misty Norwegian fjord at dawn, mountain peaks emerging through low fog, a single distant fishing boat leaves a thin wake on glassy water, the camera does a very slow forward dolly, cool blue-green palette with warm amber highlight on one peak
Atmospheric landscapes reward 1080p — the resolution preserves fog, water surface, and distant detail that compress badly at 768p. Pair a single slow camera move with a 'warm spot' element to break the cool palette.
Standard · 768p · 9:16 · 10s
Medium close-up portrait of a woman with copper-red hair sitting in a Tokyo coffee shop window seat, neon signs reflecting on wet glass behind her, she slowly looks up from her phone, eyes glistening, a small smile breaks across her face, shallow depth of field, warm interior light against cool blue rain
9:16 vertical at 768p with 10-second duration is ideal for short-form social. Hailuo handles emotional micro-beats (looking up, smile breaking, eyes glistening) better with specific verbs than generic 'happy expression'.
Unlike some models, Hailuo doesn't reward 200-word prompts. 15-30 words with a clear subject, action, camera move, and mood are usually enough. The model fills in the cinematography with reasonable defaults.
Name one dominant camera move ('slow forward dolly', 'static locked-off', 'handheld follow', '90-degree rotational pan') and let it run. Trying to combine moves in a 6-second clip usually produces a confused result.
'She slowly looks up and smiles' beats 'happy woman'. 'The runner leaps the gap and lands in a roll' beats 'parkour action'. Concrete verbs give Hailuo a motion target; adjectives leave it guessing.
768p costs less and generates faster. Use it to nail the prompt and composition. Switch to 1080p only for the final deliverable — atmospheric scenes and material-heavy product shots benefit most.
Fast mode is image-to-video — Hailuo uses your image as the first frame and writes motion onto it. The frame should already be visually rich (composition, lighting, subject pose). A blurry or generic source produces a generic clip.
Longer clips (10 seconds) are 768p-only. If you need 1080p, cap at 6s and let the editor crossfade if you need length. Don't fight the constraint.
| Setting | Values | Notes |
|---|---|---|
| Mode | Standard (text-to-video) · Fast (image-to-video) | Fast requires a starting frame; Standard generates from text alone. |
| Resolution | 768p · 1080p | 10s clips only at 768p. 1080p caps at 6s. |
| Duration | 6s · 10s | 10s requires 768p resolution. |
| Aspect ratio | 16:9 · 9:16 · 1:1 | UI hint — actual aspect comes from source frame in Fast mode. |
| Prompt optimizer | On / Off | Optional MiniMax-side prompt expansion. Try both; sometimes off produces cleaner output for already-detailed prompts. |
| Pricing | 4–20 credits per clip | Standard: 8 cr (768p 6s), 12 cr (768p 10s / 1080p 6s), 20 cr (1080p 10s — N/A). Fast: 4–7 credits. |
Standard for text-to-video from scratch — describe the scene and Hailuo composes it. Fast for image-to-video — upload a starting frame and Hailuo writes motion onto it. Fast is cheaper (4–7 credits vs 8–20) and gives you more control over the look of frame one.