Hailuo 2.3 is a multi-modal video generation model that combines text-to-video and image-to-video capabilities within a single system. It allows users to generate scenes from scratch using natural language or animate existing images with realistic motion and cinematic effects.
This version emphasizes predictable motion behavior, stronger visual continuity, and improved prompt alignment, making it more suitable for production workflows rather than experimental generation.
The model is optimized for short-form generation, making it particularly effective for digital content, advertising, and rapid prototyping workflows.
At its core, Hailuo 2.3 simplifies the video creation process by merging text-driven and image-driven generation. A single prompt can define scene composition, motion, lighting, and camera behavior, while image inputs can be expanded into dynamic sequences with minimal effort.
This flexibility makes the model equally useful for ideation and asset-based production pipelines.
One of the most noticeable improvements in Hailuo 2.3 is its handling of motion. Movements appear more natural, with better transitions between frames and more believable interactions between objects and environments.
Camera dynamics, such as pans, zooms, and tracking shots—feel smoother and more intentional, contributing to an overall cinematic quality.
The model introduces enhanced character stability across frames, which is particularly important for narrative content. Facial features, identity, and emotional tone remain consistent, even in close-up shots.
Subtle micro-expressions are handled more effectively, allowing scenes to convey emotion rather than just motion.
Hailuo 2.3 maintains a consistent visual style throughout each clip. Whether generating photorealistic footage or stylized content, the model reduces flickering and visual drift, ensuring that scenes feel cohesive from start to finish.
Hailuo 2.3 is designed to support different production needs through distinct operating modes. Instead of forcing a single balance between speed and quality, it allows users to choose based on their workflow.
The Standard mode prioritizes visual accuracy and consistency, while the Fast mode accelerates generation cycles, making it easier to experiment with prompts and variations before committing to a final render.
Hailuo 2.3 is well-suited for creating short promotional videos, product showcases, and branded visual content. Its ability to maintain object consistency and stylistic control makes it valuable for commercial production.
Content creators can quickly generate cinematic clips, animate still images, or experiment with visual storytelling formats tailored for platforms like TikTok, Instagram, and YouTube Shorts.
For filmmakers and designers, the model acts as a rapid prototyping tool. It enables quick visualization of scenes, camera angles, and narrative ideas before moving into full production.
vs Google Veo 3: Hailuo 2.3 offers superior realism in human motion and physical object interaction, with enhanced facial micro-expressions and prompt fidelity. Google Veo 3 excels in cinematic-quality video with native audio generation and excellent scene continuity. Veo 3 supports longer videos but lacks the same level of fine-grained physical realism as Hailuo 2.3.
vs Sora 2: Sora 2 targets ultra-high-resolution (up to 4K) video and longer durations (up to 60 seconds), focusing on storytelling and scene continuity. Hailuo 2.3 emphasizes physical accuracy and prompt reactivity in shorter (6-10 second) videos at Full HD. Sora 2 is better for long narrative content; Hailuo 2.3 excels in microexpression and real-time physics detail.
vs Runway Gen-4: Runway Gen-4 balances multi-scene consistency and stylized content generation suitable for creative professionals. Hailuo 2.3 outperforms in physical realism and detailed object/character interaction but offers shorter clip duration and fewer stylization options. Runway is preferred for artistic, multi-scene edits; Hailuo is ideal for photorealistic, physics-driven animation.
vs Kling 2.1: Kling 2.1 offers photorealistic video with advanced lip-syncing and extended shot capabilities targeting brand and marketing content. Hailuo 2.3 delivers enhanced micro-expressions and physical motion fidelity but supports shorter videos and less emphasis on lip-sync. Kling 2.1 is best for dialogue-heavy, branded videos; Hailuo 2.3 excels in dynamic scene and object physics.
import requests, time headers = {"Authorization": "Bearer " + AIMLAPI_KEY} job = requests.post( "https://api.aimlapi.com/v2/video/generations", headers=headers, json={ "model": "minimax/hailuo-2.3", "prompt": "A serene timelapse of clouds over a mountain range" }, ).json() gid = job["id"] while True: res = requests.get(f"https://api.aimlapi.com/v2/video/generations?generation_id={gid}", headers=headers).json() if res.get("status") in ("completed", "error"): break time.sleep(5) print(res["video"]["url"])
const headers = { Authorization: `Bearer ${process.env.AIMLAPI_KEY}`, "Content-Type": "application/json", }; const job = await (await fetch("https://api.aimlapi.com/v2/video/generations", { method: "POST", headers, body: JSON.stringify({ "model": "minimax/hailuo-2.3", "prompt": "A serene timelapse of clouds over a mountain range" }), })).json(); let res; do { await new Promise((r) => setTimeout(r, 5000)); res = await (await fetch(`https://api.aimlapi.com/v2/video/generations?generation_id=${job.id}`, { headers })).json(); } while (!["completed", "error"].includes(res.status)); console.log(res.video.url);
# submit the generation — the response contains the job "id" curl -X POST https://api.aimlapi.com/v2/video/generations \ -H "Authorization: Bearer $AIMLAPI_KEY" \ -H "Content-Type: application/json" \ -d '{"model":"minimax/hailuo-2.3","prompt":"A serene timelapse of clouds over a mountain range"}' # then poll until status is "completed" — video.url holds the result curl "https://api.aimlapi.com/v2/video/generations?generation_id={id}" -H "Authorization: Bearer $AIMLAPI_KEY"
OpenAI-compatible — swap the base URL and it works with your existing SDK.
| Type | Price |
|---|---|
| Output | |
| Benchmark | Score | What it measures | Source | Retrieved |
|---|---|---|---|---|
| Image-to-Video Arena | 1255 | Human preference Elo from blind pairwise comparisons of videos animated from the same input image, measured independently by Artificial Analysis | Source | September 28, 2026 |
| Text-to-Video Arena | 1176 | Human preference Elo from blind pairwise comparisons of generated videos, measured independently by Artificial Analysis | Source | September 28, 2026 |
| Model | Input | Output | Context | Best for |
|---|---|---|---|---|
Hailuo 2.3 This page | Video generation |
Hailuo 2.3 takes image, text as input and returns video.
Hailuo 2.3 is priced at $0.728 / 10 sec.
Hailuo 2.3 is billed per generation — a fixed charge per output rather than by prompt length.
Yes, Hailuo 2.3 accepts image input as well as a text prompt.
Hailuo 2.3 was built by MiniMax.
Send a request to v2/video/generations with minimax/hailuo-2.3 as the model id.
Yes. Hailuo 2.3 is served through AI/ML API, so the same key and endpoint format used for other models applies.
-min.webp){kind=small}
