Elasticsearch has released version 9.0 and 8.18, which includes capabilities previously available to Elastic Cloud Serverless users. The new releases enable fully managed Elasticsearch on AWS, Azure, and GCP. One of the key features is the ability to perform semantic searches using the "semantic_text" type in mappings. By default, the ELSER model is used, but users can specify a custom inference endpoint, such as Jina AI's latest embedding model. A Jupyter notebook is available to try out the feature. The semantic search capability allows for natural language queries, and results can be highlighted to show relevant fragments. The ES|QL query language is also supported, and a blog post provides more information on using ES|QL for joins. The releases provide users with flexibility in choosing their inference endpoint and query language. Overall, the new versions of Elasticsearch offer enhanced search capabilities and flexibility.