New Integrations

• Redis: implement RedisChatMemoryStore by @zambrinf in https://github.com/langchain4j/langchain4j/pull/1358
• OVHcloud: integrate embedding models by @philippart-s in https://github.com/langchain4j/langchain4j/pull/1355

Notable Changes

• Google Gemini: support audio, video and PDF inputs by @glaforge in https://github.com/langchain4j/langchain4j/pull/1464
• Ollama: migrate to Jackson by @Martin7-1 in https://github.com/langchain4j/langchain4j/pull/1072
• Amazon Bedrock: support Titan embedding model V2 (amazon.titan-embed-text-v2:0) by @staleks in https://github.com/langchain4j/langchain4j/pull/1487
• Make enum values work with @Description annotation by @PrimosK in https://github.com/langchain4j/langchain4j/pull/1398
• Make it possible to return list of enum values annotated with @Description annotation by @PrimosK in https://github.com/langchain4j/langchain4j/pull/1410
• Chroma: support embedding removal by @Heezer in https://github.com/langchain4j/langchain4j/pull/1457
• Pinecone: support storing metadata and embedding removal by @Martin7-1 in https://github.com/langchain4j/langchain4j/pull/1400

Breaking Changes

• Fix split packages by @langchain4j in https://github.com/langchain4j/langchain4j/pull/1433

We had to make some breaking changes to fix split package issues. Most of the changes should not affect your code; however, this one most probably will: All in-process ONNX model classes moved into their own unique packages

• AllMiniLmL6V2EmbeddingModel moved from dev.langchain4j.model.embedding into dev.langchain4j.model.embedding.onnx.allminilml6v2 package
• AllMiniLmL6V2QuantizedEmbeddingModel moved from dev.langchain4j.model.embedding into dev.langchain4j.model.embedding.onnx.allminilml6v2q package
• OnnxEmbeddingModel moved from dev.langchain4j.model.embedding into dev.langchain4j.model.embedding.onnx package
• etc

Other Changes

• Fix the incorrect endpoint for ERNIE-Speed-128K. by @likaiqiang in https://github.com/langchain4j/langchain4j/pull/1426
• Sanitize messages before sending to Qwen models by @jiangsier-xyz in https://github.com/langchain4j/langchain4j/pull/1423
• [Doc] Refactor the Azure documentation by @agoncal in https://github.com/langchain4j/langchain4j/pull/1431
• Add Creating AzureOpenAiChatModel with Azure Credentials to the doc by @agoncal in https://github.com/langchain4j/langchain4j/pull/1435
• Fix zhipuai SystemMessage role field is empty by @1402564807 in https://github.com/langchain4j/langchain4j/pull/1438
• Update the Vertex AI and Gemini SDK libraries by @glaforge in https://github.com/langchain4j/langchain4j/pull/1443
• #725: PR 1 / 4: Addresses EnumOutputParser not being called by @PrimosK in https://github.com/langchain4j/langchain4j/pull/1389
• #725: PR 2 / 4: Addresses EnumOutputParser not working properly in some r… by @PrimosK in https://github.com/langchain4j/langchain4j/pull/1392
• EmbeddingModelTextClassifier to embed all samples per labels at once by @glaforge in https://github.com/langchain4j/langchain4j/pull/1448
• Fix a enum serialization issue leading to stackoverflow when creating schemas from classes by @glaforge in https://github.com/langchain4j/langchain4j/pull/1450
• Feat(#1383): mutualise EmbeddingMatches handling by @fb33 in https://github.com/langchain4j/langchain4j/pull/1384
• automatically convert List<String> into String[] by @humcqc in https://github.com/langchain4j/langchain4j/pull/1394
• CohereEmbeddingModel: implement batching by @Apetun in https://github.com/langchain4j/langchain4j/pull/1445
• add glm-4v model by @1402564807 in https://github.com/langchain4j/langchain4j/pull/1469
• Fix ollama client missing path issue by @pangzixiang in https://github.com/langchain4j/langchain4j/pull/1456
• Bump com.azure:azure-identity from 1.13.0 to 1.13.1 by @jdubois in https://github.com/langchain4j/langchain4j/pull/1478
• re #725: PR 5: Updated documentation. by @PrimosK in https://github.com/langchain4j/langchain4j/pull/1473
• Update testcontainers version to 1.20.0 by @eddumelendez in https://github.com/langchain4j/langchain4j/pull/1488
• Bump com.knuddels:jtokkit from 1.0.0 to 1.1.0 by @dafriz in https://github.com/langchain4j/langchain4j/pull/1499
• DefaultRetrievalAugmentor: process in the same thread when single query and single retriever, otherwise use executor and cache threads for 1 second by @langchain4j in https://github.com/langchain4j/langchain4j/pull/1479
• Release 0.33.0 by @langchain4j in https://github.com/langchain4j/langchain4j/pull/1514

New Contributors

• @likaiqiang made their first contribution in https://github.com/langchain4j/langchain4j/pull/1426
• @PrimosK made their first contribution in https://github.com/langchain4j/langchain4j/pull/1389
• @Apetun made their first contribution in https://github.com/langchain4j/langchain4j/pull/1445
• @pangzixiang made their first contribution in https://github.com/langchain4j/langchain4j/pull/1456
• @dafriz made their first contribution in https://github.com/langchain4j/langchain4j/pull/1499
• @staleks made their first contribution in https://github.com/langchain4j/langchain4j/pull/1487
• @philippart-s made their first contribution in https://github.com/langchain4j/langchain4j/pull/1355

Full Changelog: https://github.com/langchain4j/langchain4j/compare/0.32.0...0.33.0