What's Changed
- e2e: disable failing tests and environments by @mikebonnet in #2791
- e2e: fix registry push test by using --network=host by @olliewalsh in #2793
- convert: switch to using llama-convert-hf-to-gguf by @mikebonnet in #2786
- Update react monorepo to v19.2.7 by @renovate[bot] in #2790
- Update CI updates by @renovate[bot] in #2789
- test: use simpler math prompt in sandbox stdin test by @bmahabirbu in #2794
- Show container image pull progress unless --quiet is set by @olliewalsh in #2780
- Show shortname column in ramalama list output by @rhatdan in #2766
- huggingface: use the HF_TOKEN env var if it's defined by @mikebonnet in #2797
- feat: add multi-model router mode for ramalama serve by @bmahabirbu in #2563
- Relax ollama deprecation now that is it pivoting back to llama.cpp by @olliewalsh in #2803
- feat: add --caption-images flag for VLM-based image description in RAG by @bmahabirbu in #2654
- Bump llama.cpp to b9701 by @olliewalsh in #2805
- Add ncmoe as parameter by @mkristian in #2778
- Update CI updates (major) by @renovate[bot] in #2809
- konflux: Pass HF_TOKEN from Kubernetes secret to e2e test containers by @mikebonnet in #2810
- Update all dependencies in the -rag and -tools images to their latest versions by @mikebonnet in #2807
- Bump version to v0.23.0 by @olliewalsh in #2814
Full Changelog: v0.22.0...v0.23.0