What's Changed
✨ New Features
- feat: Add multi lang support for OCR by @mfortman11 in #2393
- feat: character handling fix by @mfortman11 in #2453
- feat: Improve RHOAI model provider error handling and resiliency by @mpawlow in #2419
- feat: Design update cardless by @mfortman11 in #2415
- feat: route Azure AI Foundry over its OpenAI-compatible v1 surface by @ricofurtado in #2445
- feat: move all frontend calls to axios, add base path on frontend build by @lucaseduoli in #2466
- feat: enable watsonx vision enabled models in vlm settings by @mfortman11 in #2459
- feat: Implement model provider discovery for Red Hat OpenShift API by @mpawlow in #2482
- feat: Enable Azure AI Models in Saas and CPD by @edwinjosechittilappilly in #2513
- feat: Add Langfuse tracing by @mfortman11 in #2510
- feat: Improve Red Hat OpenShift API model provider settings UI / UX by @mpawlow in #2525
- feat: add retrieval evaluation harness (#2478) by @Thenathanb in #2498
🐛 Bug Fixes
- fix: validate chunk settings before individual-file ingest by @Thenathanb in #2433
- fix: keep IME confirmation Enter from sending the chat On macOS, Safari, Chrome, and Edge fire Enter when a Japanese candidate is confirmed. That key was treated as send, so the message went out before the user finished typing. by @Wallgau in #2450
- fix: exclude both luna generations and their dated snapshots by @ricofurtado in #2444
- fix: surface a failed duplicate check instead of ingesting anyway by @ricofurtado in #2434
- fix: make a duplicate check answerable from the server logs by @ricofurtado in #2461
- fix: azure openAI button issue by @VisveshJ in #2442
- fix: pass account name explicitly for Azure Blob account key auth by @Thenathanb in #2455
- fix: embed through the gateway for every provider, not just aliased ones by @ricofurtado in #2443
- fix: a status with no entry no longer takes down the knowledge view by @ricofurtado in #2468
- fix: reduce flakiness of tests and fix failing e2e tests by @lucaseduoli in #2481
- fix: an upload can no longer replace a shared document without the permission to delete one by @ricofurtado in #2479
- fix: clear exception to aggrid render module by @VisveshJ in #2489
- fix: Address CVE-2026-86140 by @mfortman11 in #2497
- fix: Address CVE-2026-74860 by @mfortman11 in #2496
- fix: cve for openrag-frontend & openrag-backend remeditations by @jason-dong03 in #2499
- fix: address pyjwt and authlib CVEs by @VisveshJ in #2500
- fix: Address Js CVE fixes by @mfortman11 in #2501
- fix: Status icon fix by @mfortman11 in #2452
- fix: ignore axios cancellations in nudges and conversations queries by @VisveshJ in #2506
- fix: Docling VLM endpoints and separators for both Azure providers by @ricofurtado in #2446
- fix: report an over-length embedding input as what it is by @ricofurtado in #2474
- fix: embed an over-length input in pieces instead of failing it by @ricofurtado in #2477
- fix: correct Azure AI Foundry resource health checks by @edwinjosechittilappilly in #2514
- fix: backend CVE's by @VisveshJ in #2518
- fix: cpd Ingestion fails with '413 Request Entity Too Large' for files of size 1MB and greater by @Wallgau in #2470
- fix: apply the search score threshold before DLS adds its own score by @ricofurtado in #2520
- fix: keep interrupted Langflow streams owned and resumable (port #2527 to main) by @edwinjosechittilappilly in #2531
- fix: change litellm tests in order to change deprecated model by @lucaseduoli in #2543
- fix: Detect password-protected PDFs when docling omits the PDFium reason by @mpawlow in #2540
- fix: Upgrade Next.js to 16.4.3 for GHSA-2464-fhr4-w9xw, GHSA-673q-mjm6-r5xr, GHSA-p3m5-59mp-jv5r by @mpawlow in #2547
- fix: added modified time as tag on langflow flows so that updated versions are always prompted to upgrade by @lucaseduoli in #2551
- fix: split oversized ingest chunks by @Wallgau in #2549
📝 Documentation Updates
✅ Tests
- test: live validation matrix for Azure AI Foundry, ahead of the flip by @ricofurtado in #2447
Others
- chore: ui simplification / QoL Fixes by @jason-dong03 in #2423
- chore: upgrade ci node versions by @mfortman11 in #2494
- chore: Upgrade node to avoid EoLs by @mfortman11 in #2493
- chore: update openai hardpin by @mfortman11 in #2495
- chore: Python image, versioning and EoL fixes by @mfortman11 in #2502
- chore: upgrade to tailwind v4 by @mfortman11 in #2504
- chore: Address Overdue cves by @mfortman11 in #2517
- chore: Address CVE-2026-59951 and GHSA-wq5f-xc86-pv6w by @mfortman11 in #2530
- chore: update docs with current way of contributing to OpenRAG by @lucaseduoli in #2539
Full Changelog: v0.8.0...v0.8.1