Alibaba announced Qwen3.8, a 2.4-trillion-parameter model slated for open-weight release. A preview version is available through Alibaba's Token Plan, Qoder, and QoderWork.

The Chinese tech giant is simultaneously open-sourcing the software stack for its Zhenwu series of AI chips, targeting Nvidia's CUDA ecosystem. The move aims to lower migration barriers for developers currently locked into CUDA, as China seeks AI independence beyond just building alternative hardware.

Moonshot AI launched Kimi Code CLI, an AI coding agent that runs in the terminal. The tool can read and edit code, run shell commands, search files, and fetch web pages while choosing next steps based on feedback.

The company's Kimi K3 model has received unexpectedly high demand, forcing Moonshot to temporarily pause new subscriptions and prioritize compute for current members. The startup plans to list on the Hong Kong Stock Exchange within six months.

Meanwhile, Apple has sent legal letters to dozens of former Apple employees now working at OpenAI. The letters instruct employees to preserve documents related to Apple's trade secret lawsuit, which claims OpenAI recruited key engineers and benefited from proprietary designs. More than 400 former Apple employees now work at OpenAI.

Netflix detailed how it built its LLM serving stack within existing production infrastructure. The company covered engine selection, model packaging, API design, and deployment strategy trade-offs under real workloads.

Anthropic announced Claude Fable 5 will be included in all Max and Team Premium plans at 50% of limits starting July 20. Pro and Team Standard users will receive a one-time $100 credit to access Fable via usage credits.

Google is preparing to introduce Gemini Live to desktop and web, extending real-time voice capabilities beyond mobile devices. The rollout will include Skills functionality for enhanced interaction capabilities.