--- title: Introducing FILE type: a native column type for... description: Databricks is announcing the beta of FILE type, a new native column type for storing unstructured data (documents, images, video, audio) directly in lakehouse... canonical: https://daily.dev/posts/introducing-file-type-a-native-column-type-for-multimodal-data-w9edcnr82 twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Introducing FILE type: a native column type for multimodal data | daily.dev og:description: Databricks is announcing the beta of FILE type, a new native column type for storing unstructured data (documents, images, video, audio) directly in lakehouse... og:url: https://daily.dev/posts/introducing-file-type-a-native-column-type-for-multimodal-data-w9edcnr82 og:image: https://api.daily.dev/og/posts/W9edCnr82.png og:image:alt: Introducing FILE type: a native column type for multimodal data og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Introducing FILE type: a native column type for multimodal data **[databricks](https://daily.dev/sources/databricks)** · 5 min read · 0 upvotes · 0 comments ## Summary Databricks is announcing the beta of FILE type, a new native column type for storing unstructured data (documents, images, video, audio) directly in lakehouse tables. FILE columns store lightweight references rather than raw bytes, so query performance is not degraded when working with large files. The type integrates with Unity Catalog for row- and column-level access controls and ABAC, solving the dual-permission-model problem that arises when storing file paths as strings. Lifecycle management is also unified: deleting a row deletes the associated file reference, preventing orphaned files and simplifying GDPR compliance. Databricks is contributing FILE support upstream to Apache Parquet and Delta Lake to keep the format open and portable. Built-in functions like AI_PARSE_DOCUMENT and AI_QUERY can be applied directly over FILE columns to extract structured insights for agents and analysts. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### How does Databricks FILE type handle large binary files without hurting query performance? FILE columns store a lightweight reference to the file rather than the raw bytes. The engine pulls the actual bytes only at the step that needs them, so operations that don't require the file content don't drag gigabytes through memory. This contrasts with encoding raw binary directly in the table, where every query operation carries the full binary payload. _Teams building multimodal pipelines on Databricks track FILE type updates and patterns on daily.dev._ ### How does Databricks FILE type handle access control compared to storing file paths as strings? Storing file paths as strings requires maintaining two separate permission models — one for the table and one for the object storage folder. FILE columns are integrated into Unity Catalog, so they are governed by the same row-level, column-level, and attribute-based access controls (ABAC) as structured columns. A row filter that restricts a FILE column also restricts the underlying file, not just the path string. _Architects choosing between path-string and FILE-column approaches for governed data lakes follow these decisions on daily.dev._ ### Does Databricks FILE type work with open formats like Parquet and Delta Lake? Yes. Databricks is contributing FILE type support directly into Apache Parquet and Delta Lake as open-source changes, with commits already visible in both the parquet-format and delta-io/delta repositories. This means the format is not proprietary to Databricks — the broader ecosystem can adopt it, keeping multimodal data portable across vendors and model providers. _Engineers evaluating vendor lock-in risk for lakehouse multimodal storage follow open-format developments on daily.dev._ ## Similar posts on daily.dev - [Parquet Files Explained: Why Every Data Engineer should Use](https://daily.dev/posts/parquet-files-explained-why-every-data-engineer-should-use-mz7jytdxj) · Medium · 1 upvotes · 0 comments - [Parquet and Iceberg: An Overview — CommandPrompt Inc.](https://daily.dev/posts/parquet-and-iceberg-an-overview-commandprompt-inc--pv4vposch) · Planet PostgreSQL · 0 upvotes · 0 comments - [Ingest semi-structured data faster and more efficiently with Variant - Now Generally Available](https://daily.dev/posts/ingest-semi-structured-data-faster-and-more-efficiently-with-variant---now-generally-available-yxwphvq9m) · databricks · 0 upvotes · 0 comments - [Borderless Lakehouse cross-cloud caching and connections](https://daily.dev/posts/borderless-lakehouse-cross-cloud-caching-and-connections-xctvphcsx) · Google Cloud · 0 upvotes · 0 comments - [Bringing DuckLake to DataFusion](https://daily.dev/posts/bringing-ducklake-to-datafusion-hcebiftiz) · DuckLake Blog · 0 upvotes · 0 comments --- Tags: [#big-data](https://daily.dev/tags/big-data), [#databricks](https://daily.dev/tags/databricks), [#unity-catalog](https://daily.dev/tags/unity-catalog) [View this post on daily.dev](https://daily.dev/posts/introducing-file-type-a-native-column-type-for-multimodal-data-w9edcnr82) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Introducing FILE type: a native column type for multimodal data","url":"https://daily.dev/posts/introducing-file-type-a-native-column-type-for-multimodal-data-w9edcnr82","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/introducing-file-type-a-native-column-type-for-multimodal-data-w9edcnr82"},"datePublished":"2026-08-07T15:58:41.368Z","dateModified":"2026-08-07T15:59:13.671Z","description":"Databricks is announcing the beta of FILE type, a new native column type for storing unstructured data (documents, images, video, audio) directly in lakehouse...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/361b1e6c9a1deaf71458241b6032fd64?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/361b1e6c9a1deaf71458241b6032fd64?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"databricks","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"databricks","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/fa7aa720f2db4d1eba826814730482c8","url":"https://daily.dev/sources/databricks"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/introducing-file-type-a-native-column-type-for-multimodal-data-w9edcnr82","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"big-data,databricks,unity-catalog","timeRequired":"PT5M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"databricks","item":"https://daily.dev/sources/databricks"},{"@type":"ListItem","position":3,"name":"Introducing FILE type: a native column type for multimodal data"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/introducing-file-type-a-native-column-type-for-multimodal-data-w9edcnr82#faq","mainEntity":[{"@type":"Question","name":"How does Databricks FILE type handle large binary files without hurting query performance?","acceptedAnswer":{"@type":"Answer","text":"FILE columns store a lightweight reference to the file rather than the raw bytes. The engine pulls the actual bytes only at the step that needs them, so operations that don't require the file content don't drag gigabytes through memory. This contrasts with encoding raw binary directly in the table, where every query operation carries the full binary payload. Teams building multimodal pipelines on Databricks track FILE type updates and patterns on daily.dev."}},{"@type":"Question","name":"How does Databricks FILE type handle access control compared to storing file paths as strings?","acceptedAnswer":{"@type":"Answer","text":"Storing file paths as strings requires maintaining two separate permission models — one for the table and one for the object storage folder. FILE columns are integrated into Unity Catalog, so they are governed by the same row-level, column-level, and attribute-based access controls (ABAC) as structured columns. A row filter that restricts a FILE column also restricts the underlying file, not just the path string. Architects choosing between path-string and FILE-column approaches for governed data lakes follow these decisions on daily.dev."}},{"@type":"Question","name":"Does Databricks FILE type work with open formats like Parquet and Delta Lake?","acceptedAnswer":{"@type":"Answer","text":"Yes. Databricks is contributing FILE type support directly into Apache Parquet and Delta Lake as open-source changes, with commits already visible in both the parquet-format and delta-io/delta repositories. This means the format is not proprietary to Databricks — the broader ecosystem can adopt it, keeping multimodal data portable across vendors and model providers. Engineers evaluating vendor lock-in risk for lakehouse multimodal storage follow open-format developments on daily.dev."}}]} ```