Единый шлюз для AI-провайдеров с OpenAI API

★ 7.9 · ai-tooling

omni-inference is a Claude Code skill that serves as the primary integration surface for AI agents, exposing OpenAI-compatible endpoints for chat completions, embeddings, image generation, audio (TTS/STT), content moderation, reranking, and the Responses API — all on a single gateway running at port 20128. Requests can be routed generically or directed to a specific provider via `/api/v1/providers/{provider}/...`, with additional compatibility layers for Anthropic Messages API and Ollama's `/api/chat` format. A WebSocket endpoint supports streaming completions and mid-flight cancellation, while a dedicated live channel on port 20129 streams dashboard events across `requests`, `combo`, and `credentials` topics with a 15-second heartbeat. Designed for developers building multi-provider AI pipelines who need one authenticated entry point instead of managing separate SDKs.