v0.23.0

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