Vidu AI

Vidu AI is an AI-powered multimodal video-generation platform developed by ShengShu AI. It allows creators, developers, and businesses to generate and edit visual content using text, images, reference materials, and other inputs. Vidu supports several video-generation workflows, including Text-to-Video, Image-to-Video, Reference-to-Video, and Start-End-to-Video. Its platform has expanded beyond standard video generation to include real-time interactive video, image generation, audio generation, digital characters, and video-editing capabilities.

Vidu’s current model ecosystem is organized around several model families. The Vidu Q3 series is the platform’s current flagship video generation family, with variants including Q3-Turbo, Q3-Pro, Q3-Mix, Q3-Ad, and Q3-Drama. These variants are designed for different production requirements, such as faster generation, balanced general-purpose creation, advertising, and drama-oriented content. The platform also provides Vidu Q2 and Q1 models for additional video-generation workflows.

One of Vidu’s central capabilities is Text-to-Video generation. Users can describe a scene using a written prompt and have the platform generate a video based on that description. The API supports configurable parameters such as duration, aspect ratio, resolution, style, and motion amplitude. This gives developers more control over the generated output when incorporating Vidu into an application or automated workflow.

Image-to-Video allows users to transform a still image into an animated video. Vidu can add movement and visual effects to a supplied image while maintaining the main visual subject. The platform also supports Start-End-to-Video, where users provide beginning and ending frames and Vidu generates the transition between them. This is useful for workflows that require greater control over how a video begins and ends.

Reference-to-Video is another important feature. Users can provide multiple reference images of a character, object, or scene and ask Vidu to generate a video while maintaining the identity and visual characteristics of those references. The official platform describes this as a way to improve subject consistency and reuse characters, props, and scenes across different generations.

Vidu has also developed specialized capabilities for animation and commercial content. Its documentation highlights strong performance for anime-style video and subject consistency, while the current model lineup includes a Q3-Ad variant designed specifically for advertising and e-commerce applications. The platform also provides templates and one-click production features for advertising, general films, AI music videos, and other content workflows.

The platform is expanding into real-time video generation. Its current S2 series includes Vidu S2 Avatar and Vidu S2 Editing. S2 Avatar supports real-time voice interaction and motion control, while S2 Editing can use references for style, clothing, characters, and backgrounds when editing incoming video streams in real time. These capabilities make Vidu broader than a conventional prompt-to-video generator.

Vidu also provides image and audio capabilities through its API. Its current model map includes Vidu Q2 Image and Vidu Q1 Image for text-to-image and reference-to-image generation. The audio section includes voice cloning, text-to-audio generation, and text-to-speech. This gives developers access to multiple media-generation capabilities through the same API ecosystem.

For businesses and developers, Vidu provides an API platform that can be integrated into applications and production workflows. Developers can create API keys, purchase credits, send HTTPS requests, select models, specify generation parameters, and retrieve generated results. The platform is designed for both individual development and enterprise-scale deployments, with dedicated concurrency, quotas, private deployment, and model fine-tuning available through its enterprise offering.

Vidu’s API uses a credit-based payment system. Credits are purchased and consumed according to the type of generation, model, resolution, and duration. The current standard credit price is $0.005 per credit, although the actual cost of a generation varies substantially between models. For example, current Q3 Pro text/image/start-end-to-video generation ranges from $0.045 per second at 540P to $0.12 per second at 1080P, while Q3 Turbo starts lower.

The service also provides free credits for users of the consumer Vidu platform. This gives users an opportunity to experiment with AI video generation before purchasing additional credits. For API users, credits are purchased through the billing dashboard, and the platform supports single credit purchases between $10 and $10,000.

Vidu was commercially launched in July 2024 by ShengShu Technology. The company subsequently released Vidu 1.5 in November 2024 and Vidu 2.0 in January 2025, followed by newer Q-series and S-series models. The company has continued expanding Vidu from a video generator into a broader multimodal generation platform for creators, developers, and enterprises.

Pricing

Vidu uses a credit-based, usage-based pricing model. The consumer platform provides free credits, while additional usage is paid for through credits. The API currently prices credits at $0.005 per credit, with the exact generation cost depending on the selected model, resolution, duration, and generation type.

Current examples for Vidu Q3 include:

• Q3 Pro 540P: $0.045 per second
• Q3 Pro 720P: $0.10 per second
• Q3 Pro 1080P: $0.12 per second
• Q3 Turbo 540P: $0.035 per second
• Q3 Turbo 720P: $0.055 per second
• Q3 Turbo 1080P: $0.065 per second
• Q3 reference-to-video: from $0.035 per second depending on model and resolution

Off-peak generation can reduce eligible generation costs by approximately 50%.

Pricing Note

Vidu does not use one fixed price for every generation. Costs vary according to the model, resolution, duration, task type, and whether off-peak pricing is available. The platform’s pricing documentation notes that the displayed pricing should be checked for current rates because model pricing and available packages can change.

Review

Vidu AI has developed from a conventional AI video generator into a broader multimodal creation platform. Its combination of text-to-video, image-to-video, reference-to-video, start/end-frame generation, audio, image generation, digital characters, and real-time models gives users several ways to approach video production.

Subject consistency is one of Vidu’s major areas of focus. Reference-to-Video allows creators to provide several images of a character or object and use those references across generated scenes. This can be particularly useful for animation, product visualization, advertising, and storytelling workflows where maintaining the identity of the main subject is important.

Vidu also has a strong developer offering. Its API provides access to multiple video, image, audio, and streaming models from a unified platform. Developers can select different models depending on whether they need faster generation, higher-quality output, advertising-focused generation, drama-oriented scenes, image generation, voice cloning, or real-time interaction.

The main consideration is the credit-based pricing structure. Because different models and resolutions consume different numbers of credits, the cost per finished video can vary considerably. High-volume users therefore need to monitor credit consumption carefully. The open platform also has concurrency limits associated with credit packages, while enterprise customers can obtain higher concurrency and customized arrangements.

For creators, studios, marketers, developers, and businesses looking for a video platform with multiple generation modes, reference control, commercial-content workflows, and API access, Vidu provides a broad set of current video and multimodal generation capabilities.