Process files, images, audio, and video natively with SQL and Python. Connect lake formats, vector databases, and analytics systems, and scale from your machine to Ray clusters.
import vane con = vane.connect() images = con.sql(""" SELECT image_file( 'quickstart-images/large.png' ) AS source """) images.select(""" source, image_file_metadata(source).width AS width, resize( decode_image_file(source, 'RGB'), 224, 224 ) AS image """).show()
FILE, IMAGEFILE, AUDIOFILE, and VIDEOFILE represent file references. Decode and transform media as IMAGE pixels and TENSOR values with built-in operators.
Python media extras provide the default backend; native_media acceleration is optional.
Explore multimodal types →Lake and columnar providers: Iceberg, Paimon, DuckLake, Lance, and Vortex. Distributed upserts to Milvus and Qdrant; Arrow batch writes to Doris.
Lake and columnar providers are installed separately.
Explore data connections →From multimodal model training to enterprise data pipelines, real-world AI runs on diverse data. Pick the pipeline that matches your workload.
Read typed media and lake tables, then filter, label, and deduplicate data for multimodal model training.
Explore →Process business files with SQL and model UDFs, then write structured records and embeddings to retrieval and analytics systems.
Explore →Benchmarking multimodal AI pipelines across audio, video, document, and image workloads.
36 CPU cores · 64 GB RAM2080 Ti (modified VRAM) · 22 GB GPU memory
This single-node evaluation adapts the workload design from the Ray Data multimodal AI benchmark. Daft's OOMs on the image and audio workloads may reflect the test machine's limited memory, while the older 2080 Ti GPU also limits throughput. Given current hardware and compute-budget constraints, we have not reproduced the original benchmark's full cluster-scale setup; these results apply only to the recorded single-node environment.
Unifies data, models, and agents — enabling continuous learning and scalable execution from local devices to Ray clusters.
pip install vane-ai
Start from the docs examples and adapt the pipeline to your data.
Use the docs examples and llms.txt files to wire Vane into your stack.