Changelog

All notable changes to the Woven City AI Vision Engine are documented on this page.

Release Date: February 24, 2026

Features:

VLLMClient API for vision-language model inference
Image-based chat functionality (image_chat)
Video-based chat functionality (video_chat)
RGB image array input support (rgb_image_chat, rgb_images_chat)
Multi-turn conversation support with chat history
Multiple sampling methods: duration, fps, middle, random
Sampling information output with return_sampling_info parameter
Generation configuration support (temperature, top_k, top_p, max_new_tokens, etc.)

← Previous: FAQ