2.19.0

New features & Performance improvements

• Re-introduced support for Languagebind Models. You can now create indexes that support audio and video using these models. Environment variables for video and audio file size limits (MARQO_MAX_SEARCH_VIDEO_AUDIO_FILE_SIZE and MARQO_MAX_ADD_DOCS_VIDEO_AUDIO_FILE_SIZE) have been moved to the API layer (#1188).
• Significant performance improvements with upgrade to Pydantic V2. This change results in up to 6x increase in search result post-processing performance and 25% increase in end-to-end throughput. String array field handling and field map property evaluation have also been improved (#1196).
• Add new endpoint POST indexes/{index_name}/documents/get-batch to retrieve documents via POST call (#1190).

Bug fixes and minor changes

• Fix bug caused by race condition where concurrent add_documents and update_documents calls temporarily cause an error in search due to inconsistent return values (#1194).
• Update telemetry key of image downloading from image_download to media_download.{modality} (#1206).
• Fix bug where ffmpeg processes video files as audio files by adding modality check in streaming media preprocessor (#1209)
• Improve differentiation between video and image in streaming media preprocessor by searching for _pipe in format name (13a8228)

Contributor shout-outs

• Shoutouts to our valuable 4.8k stargazers!
• Thanks a lot for the discussion and suggestions in our community. We love to hear your thoughts and requests. Join our Slack channel and forum now.