LLM4AD_Next v1.0.0

Highlights

LLM4AD_Next v1.0.0 introduces the first public llm4ad-next package release, versioned Docker Hub deployment images, the online demo domain, and the latest AI Build and Web UI improvements.

Online Trial

Try LLM4AD_Next online: https://llm4ad-next.cn/

Package and Deployment

• Published the Python SDK under the PyPI distribution name llm4ad-next.
• Kept the Python import package and CLI stable as llm4ad.
• Added release automation for versioned Docker Hub images:
  • docker.io/noah2012/backend:v1.0.0
  • docker.io/noah2012/frontend:v1.0.0
  • docker.io/noah2012/llm4ad-task-runner:v1.0.0
• Added PyPI Trusted Publisher based publishing through GitHub Actions.
• Kept Docker latest publishing separate in the existing main-branch Docker workflow.

Core Updates

• Added AI Build Beta support with an AgentScope-based task-package build agent.
• Added the llm4ad chatv2 agent build command and related documentation.
• Enabled Python 3.12 as the required runtime baseline.
• Enabled the AI Build / chatv2 path out of the box.
• Added a News section backed by GitHub Wiki content with per-language cache support and manual refresh.
• Switched online trial and documentation references from raw IP addresses to llm4ad-next.cn.
• Added Docker Hub centered deployment defaults and versioned release workflow.
• Added star history chart automation and live QR update automation.
• Regenerated frontend API clients for embedding provider and beta endpoint compatibility.
• Improved embedding provider configuration and surfaced embedding scheduling failures.

Installation

pip install llm4ad-next

Optional dependency groups are also available:

pip install "llm4ad-next[providers,eval]"
pip install "llm4ad-next[all]"

Docker Deployment

cd docker
TAG=v1.0.0 ./start.sh start