> ## Documentation Index
> Fetch the complete documentation index at: https://novita.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# September 28, 2025 Product Updates

## **Model APIs Updates**

### **New Features**

* **qwen/qwen3-vl-235b-a22b-thinking Model Release**

  Qwen3-VL—the most powerful vision-language model in the Qwen series to date. Novita offers world-first support!
* **qwen/qwen3-vl-235b-a22b-instruct Model Release**

  Qwen3-VL—the most powerful vision-language model in the Qwen series to date. Novita offers world-first support!
* **minimax-hailuo-02 First-last Frame**

  First-Last Frame is a video generation technology introduced by Hailuo AI. By specifying the start and end images (the first and last frames), it automatically generates dynamic transition videos in between. The technology supports complex instruction execution and delivers cinematic-quality effects, making it suitable for applications such as e-commerce advertising, one-click outfit changes, and more.
* **Seedance V1 Lite/Pro**

  Seedance 1.0 is the latest video generation foundational model launched by ByteDance's Doubao AI team, featuring both lite and pro versions. Its capabilities and pricing are fully aligned with the official website.

  Seedance V1 Lite Text to Video

  Seedance V1 Lite Image to Video

  **Seedance V1 Pro Text to Video**

  **Seedance V1 Pro Image to Video**

***

## **GPUs Updates**

### **New Features**

* **Subscribtion Instance Auto-Renewal Feature**

  Supports setting up automatic renewal for Subscribtion Instances.
* **Creating instances and templates support the function of specifying the minimum CUDA version**

  Creating instances and templates support the function of specifying the minimum CUDA version.