Models
Tencent previews Hy Image3.5 for professional visual creation
Tencent's Hunyuan site lists a September 21 preview announcement focused on image quality and professional creative tasks.
Models.
The capabilities, trade-offs and people behind China’s AI models.
70 storiesPage 9 of 12
Models
StepFun's Step 5 Preview targets long-running agent workflows
The documented model supports text, image and video input with a million-token context window.
Models
Qwen adds a long-context multimodal Flash model
Qwen3.8-Omni-Flash handles text, images, audio and video, with a reported context capacity of one million tokens.
Models
AWS brings Moonshot's Kimi K3 to Amazon Bedrock
AWS announced Bedrock availability on September 18, adding another distribution channel for the Chinese model developer.
Models
Huawei opens XiaoYi Work beta with Windows access and HarmonyOS reservations
Huawei's work assistant connects mobile task management with local or cloud computers; current downloads and beta access vary by platform.
Models
vivo previews BlueCode and explores a larger on-device model
vivo demonstrated a phone-oriented coding agent and discussed work toward deploying a 30-billion-parameter mixture-of-experts model on devices.
