Skip to content

v0.1.0 - Initial public release

Choose a tag to compare

@aaa17001 aaa17001 released this 31 Aug 15:12
· 1 commit to main since this release

Initial public release of Codex Multi-Model Workers.

  • MoE4All local GGUF worker management on loopback
  • GLM and DeepSeek official APIs or custom compatible endpoints
  • Generic OpenAI-compatible Chat Completions providers
  • Advisory routing with Codex retained as controller
  • Safe VS Code chatLanguageModels.json backup and merge
  • Self-contained MCP bundle; end users do not run npm install
  • API keys are read only from environment variables