# Proposing OrcaRouter Native Integration for Verba (OpenAI-compatible LLM gateway) **URL:** \ **Category:** Support\ **Tags:** developer-experience\ **Created:** [May 27, 2026, 1:08pm UTC](https://forum.weaviate.io/t/proposing-orcarouter-native-integration-for-verba-openai-compatible-llm-gateway/22460 "2026-05-27T13:08:52Z")\ **Posts on this page:** 1\ **Page:** 1
**Author:** ![Song\_Jinhao](https://yyz1.discourse-cdn.com/flex027/user_avatar/forum.weaviate.io/song_jinhao/32/3748_2.png) [@Song\_Jinhao](https://forum.weaviate.io/u/Song_Jinhao)\ **Post date:** [May 27, 2026, 1:08pm UTC](https://forum.weaviate.io/t/proposing-orcarouter-native-integration-for-verba-openai-compatible-llm-gateway/22460/1 "2026-05-27T13:08:52Z")
Hi Weaviate & Verba team, I’ve opened a PR adding **OrcaRouter** as a native LLM provider in Verba: [https://github.com/weaviate/Verba/pull/470](https://github.com/weaviate/Verba/pull/470) #### About OrcaRouter OrcaRouter is an OpenAI-compatible API gateway that fronts 160+ upstream models (OpenAI, Anthropic, DeepSeek, Gemini, etc.) at upstream cost — no markup. It also offers an adaptive router (`orcarouter/auto`) that picks a cost-efficient upstream model per request based on prompt characteristics. #### What the PR adds - A native OrcaRouter provider option in Verba’s standard LLM selection UI - Single API key access to 160+ models, so Verba users can experiment with different LLMs across their RAG pipeline (generation, reranking, etc.) without juggling multiple provider configs - Implementation is complete and verified via local end-to-end testing against a running Verba instance #### Current status This is my first PR to the Verba repo, so the GitHub Actions workflows are currently waiting on maintainer approval to run. If someone from the team has a moment to kick those off, that’d be much appreciated 🙏 No rush on the review itself — happy to iterate based on any feedback. Thanks!