LuAI ToolsLuAITools.com
提交工具
AI Tools/AI Video
AI Video

Sora

OpenAI's text-to-video AI model that turns words into cinematic clips.

What is Sora?

Sora is a video generation AI model developed by OpenAI, first unveiled in February 2024. It is built on a diffusion model and Transformer architecture—starting from static noise and gradually "denoising" step by step to produce clear video frames. What sets Sora apart is its ability to understand not just the literal meaning of your text prompts, but also how objects exist and move in the physical world. Sora 2, released on September 30, 2025, removed the length limits of the original version, introduced synchronized audio generation, and added powerful new features like character cameos and enhanced visual quality. With Sora 2, creators, filmmakers, and marketers can generate professional-quality videos up to 25 seconds long from a single text prompt or image reference, complete with synchronized dialogue, sound effects, and music.
Sora
Sora

Global Usage

Sora's journey was a classic rise-and-fall story. When the Sora app launched in October 2025, it hit 100,000 downloads on its first day and topped the App Store charts. Monthly downloads peaked at 6.1 million in November. But the momentum didn't last—downloads fell 32% in December, then another 45% in January 2026. By February, monthly downloads had dropped to about 1.4 million. Active users peaked at around 1 million globally, then fell below 500,000. Even worse, the 30-day retention rate was only 1%, and the 60-day rate was nearly zero. Revenue was equally disappointing—total consumer revenue across iOS and Android was only about $1.4 million. On March 25, 2026, OpenAI announced it would shut down Sora. The consumer app and sora.com were discontinued on April 26, 2026, and the API will be terminated on September 24, 2026. Daily losses were estimated at around $1 million.

Key Features and Highlights

Extended video length (15-25 seconds) – A major upgrade from Sora 1's 6-second limit, enabling longer narrative sequences, multi-scene storytelling, and complete product demonstrations.

Synchronized audio generation – Sora 2 generates video and audio together, with perfectly synced dialogue that matches lip movements, environmental sound effects, background music, and multi-character conversations with realistic emotion and tone.

Character Cameo – Insert specific characters into your videos. Combined with Disney's $1 billion partnership, Sora 2 can generate videos with licensed Disney characters in custom scenes.

1080p resolution output – Full HD 1080p is now standard for all Sora 2 generations, with clear text rendering, detailed facial expressions and hand gestures, and broadcast-quality output.

Text-to-video and image-to-video – Describe your vision in natural language, or start from a still image and expand it into a dynamic video.

How to Use Sora

Sora was accessed through sora.com or the Sora iOS app, requiring a ChatGPT account login. You would enter a text description, adjust parameters like aspect ratio and resolution, and the model would generate the video—typically taking about a minute. After generation, you could use editing tools like Re-cut (trim and extend), Remix (modify based on a description), Blend (merge elements from two videos), and Loop (create seamless loops). The Sora 2 API also supported batch processing and asynchronous video generation for larger production workflows.

Usage Tips

Write prompts like a shot list—instead of "a cat," try "an orange tabby cat napping on a sunny windowsill, curtains swaying in the breeze, light shifting over time." For testing, use lower resolutions (480p costs about 4 credits/second, 720p about 16 credits/second, 1080p about 40 credits/second), then switch to 1080p for the final version. Shorter clips (4-8 seconds) tend to follow instructions more reliably—you can stitch them together in editing rather than generating one long clip.

How to Install

Sora didn't require traditional installation. You accessed it through a browser at sora.com, or downloaded the official iOS app from the App Store (compatible with iOS 16.0 or later). Android users could find it on Google Play. Developers accessed Sora through the OpenAI API.

Pricing and Plans

Sora was bundled with ChatGPT subscriptions. The Free tier (paused on January 10, 2026) offered very limited 480p, 5-second videos with watermarks. ChatGPT Plus ($20/month) provided about 50 videos per month at 720p, up to 10 seconds. ChatGPT Pro ($200/month) offered 500 videos per month at 1080p, up to 20 seconds, with priority queue processing. API pricing was pay-as-you-go: around $0.10 per second for 720p video.

Who It Was For

Sora was best suited for professional filmmakers, advertising agencies, and marketing teams needing high-quality AI video with director-level control. It was less suitable for short-form content creators needing high volume (generation was slow and costly) or users requiring strong Chinese prompt support.

Common Questions

Can Sora generate long videos? Better for short clips—4-8 seconds is the sweet spot, then stitch together. How to ensure shot consistency? Use character references—upload a character once and reuse it across videos. Does it support Chinese prompts? Chinese understanding wasn't as precise as domestic models—English prompts worked better. Are generated videos copyrighted? OpenAI had strict content filters and refused prompts involving public figures, brands, and sensitive content. Is Sora still available? The consumer app and website closed on April 26, 2026; the API ends on September 24, 2026.

Comments