​
Model APIs Updates

​
New Features

  • Batch Inference API for Large Language Models Our Batch Inference API enables asynchronous processing of high-volume inference requests while maintaining full compatibility with the OpenAI API standard. This cost-efficient solution is ideal when immediate results aren’t required, offering significantly higher rate limits than online calls while guaranteeing delivery within a reasonable 48-hour window. Perfect for: • Evaluation and data analysis at scale • Large dataset classification • Offline document summarization generation Pricing: Available at 50% off standard model inference API rates. Updated time: August 13, 2025
  • New Model Available: Vidu Q1 Vidu Q1 is a multimodal video generation model globally launched by Shengshu Technology in April 2025. It features direct output capability at 1080P resolution and AI sound synchronization functionality. The model achieves state-of-the-art (SOTA) performance in video quality and semantic consistency metrics on VBench-1.0, as well as in commonsense reasoning and physical understanding dimensions on VBench-2.0. Additionally, it ranks first in both anime and realistic style categories on the SuperCLUE text-to-video leaderboard in China. Discover the power of Vidu Q1 on Novita by exploring its text-to-video, image-to-video, start-end-to-video, and reference-to-video generation capabilities—and bring your creative ideas to life. Updated time: August 14, 2025
  • New Model Available: Vidu 2.0 Vidu 2.0 is a multimodal video generation model globally launched by Shengshu Technology in January 2025. It can generate a single 4-second video clip at 512P resolution within 10 seconds, delivering more consistent styles and themes, smoother transitions between the first and last frames, and producing four materials in one go—doubling efficiency. Discover the efficiency of Vidu 2.0 on this platform by exploring its image-to-video, start-end-to-video, and reference-to-video generation capabilities—and see how effortless video creation can be. Updated time: August 14, 2025