ComfyUI 0.45.0 adds partner nodes for Kling, OpenAI and ByteDance models

The additions cover Pika H3 Flex along with Pika soundtrack, SFX and music nodes; Kling 4.0 Flash with an original-audio option; OpenAI GPT Image 2.6 models plus GPT Image V3, GIF and SVG nodes; ByteDance Seedream 5.5 and 4K Seedance 2.5; Gemini Omni Flash 1.2 in the Omni video node; a Tencent HY LLM node; and Breeze TTS 2 speech nodes. The release also adds local support for MiniMax-H3-image, WeVisOne-VL and RelayVSR, video input for Qwen 3.x, memory-adaptive slicing for the SeedVR2 VAE, and Qwen 3.x speed work. Local web SSO and partner node tokens are now enabled by default.