# Databricks Databricks has expanded its platform with the addition of Moonshot AI's Kimi K3 model via Unity AI Gateway, the general availability of Variant for semi-structured data ingestion, the completion of its Panther acquisition, and Lakehouse//RT for millisecond query speeds. Status: provisional Last verified: 2026-08-27T19:33:02.000Z ## Evidence - (single_source) Variant is broadly integrated into the Databricks platform for AI workloads including AI Functions. - supports https://databricks.com/blog/ingest-semi-structured-data-faster-and-more-efficiently-variant-now-generally-available > Variant is broadly integrated into the Databr... AI workloads (Agent... icks, AI Functions) - (single_source) Kimi K3 from Moonshot AI is now available on Databricks through Unity AI Gateway. - supports https://databricks.com/blog/kimi-k3-moonshot-ai-now-available-databricks-through-unity-ai-gateway > Kimi K3 from Moonshot AI, the most advanced open-weight model, is available on Databricks. - (single_source) Kimi K3 is initially hosted by Databricks in the US, with more regions planned. - supports https://databricks.com/blog/kimi-k3-moonshot-ai-now-available-databricks-through-unity-ai-gateway > Kimi K3 is initially hosted by Databricks in the US, with more regions planned. - (single_source) Variant seamlessly handles schema changes so teams can ingest data without the need to update pipelines. - supports https://databricks.com/blog/ingest-semi-structured-data-faster-and-more-efficiently-variant-now-generally-available > Variant seamlessly handles schema changes, so teams can ingest data without the need to update pipelines - (single_source) Lakehouse//RT achieves response times as low as 10ms on smaller datasets and sub-100ms performance on larger ones. - supports https://databricks.com/blog/introducing-lakehousert-real-time-performance-unified-lakehouse > with response times as low as 10ms on smaller datasets and sub-100ms performance on larger ones - (single_source) Lakehouse//RT provides millisecond query speeds directly on the lakehouse without data movement. - supports https://databricks.com/blog/introducing-lakehousert-real-time-performance-unified-lakehouse > Databricks launches Lakehouse//RT, powered by new real-time engine Reyden, to provide millisecond query speeds directly on your lakehouse without data movement - (single_source) Variant is now Generally Available in Databricks for ingesting semi-structured data faster and more efficiently. - supports https://databricks.com/blog/ingest-semi-structured-data-faster-and-more-efficiently-variant-now-generally-available > We are excited to announce that Variant is now Generally Available in Databricks. - (single_source) Preview users of Lakehouse//RT have seen up to 16x better performance versus real-time serving layers. - supports https://databricks.com/blog/introducing-lakehousert-real-time-performance-unified-lakehouse > Preview users have seen up to 16x better performance vs. real-time serving layers - (single_source) Databricks has officially completed the acquisition of Panther, an AI SOC platform built for modern security operations. - supports https://databricks.com/blog/databricks-completes-acquisition-panther-accelerating-security-lakehouse-era > Databricks has officially completed the acquisition of Panther, an AI SOC platform built for modern security operations. - (single_source) Kimi K3 is described as the most advanced open-weight model. - supports https://databricks.com/blog/kimi-k3-moonshot-ai-now-available-databricks-through-unity-ai-gateway > Kimi K3 from Moonshot AI, the most advanced open-weight model, is available on Databricks. - (single_source) Models available through Unity AI Gateway are fully governed within the secure Databricks perimeter. - supports https://databricks.com/blog/kimi-k3-moonshot-ai-now-available-databricks-through-unity-ai-gateway > can build with confidence knowing the models are fully governed by Unity AI Gateway within their secure Databricks perimeter. - (single_source) Lakehouse//RT is powered by Reyden, a new engine for real-time workloads requiring immediate responsiveness at high concurrency. - supports https://databricks.com/blog/introducing-lakehousert-real-time-performance-unified-lakehouse > Lakehouse//RT is powered by Reyden, a breakthrough new engine for real-time workloads that require immediate responsiveness at high concurrency. - (single_source) Users can connect VS Code, Cursor, or a terminal directly to Databricks compute to run, debug, and scale Python and SQL workloads. - supports https://databricks.com/blog/run-debug-and-scale-databricks-workloads-your-local-ide > you can now connect VS Code, Cursor, or your terminal directly to Databricks compute. Run, debug, and scale Python and SQL workloads on real cluster infrastructure - (single_source) Attackers are leveraging automation and AI to move faster, hide within massive volumes of complex data, and launch multi-stage attacks across cloud, identity, and SaaS environments. - supports https://databricks.com/blog/databricks-completes-acquisition-panther-accelerating-security-lakehouse-era > Attackers are leveraging automation and AI to move faster, hide within massive volumes of complex data, and launch increasingly sophisticated, multi-stage attacks across cloud, identity, and SaaS environments. - (single_source) Databricks offers native access to proprietary models from Anthropic, OpenAI, and Google Gemini alongside Kimi K3. - supports https://databricks.com/blog/kimi-k3-moonshot-ai-now-available-databricks-through-unity-ai-gateway > By offering native access to Kimi K3, together with top proprietary models from Anthropic, OpenAI and Google Gemini customers have maximum choice and control - (single_source) Databricks Connect allows users to connect their local development environment to Databricks compute to remotely run Spark workloads. - supports https://databricks.com/blog/run-debug-and-scale-databricks-workloads-your-local-ide > Databricks Connect'... Connect your local development environment to Databricks compute to remotely run Spark workloads. - (single_source) Cybersecurity has fundamentally transformed into a data management and AI problem. - supports https://databricks.com/blog/databricks-completes-acquisition-panther-accelerating-security-lakehouse-era > Cybersecurity has fundamentally transformed into a data management and AI problem. - (single_source) Variant Shredding is also Generally Available as a performance optimization that uses Predictive Optimization to automatically improve query performance on Variant data. - supports https://databricks.com/blog/ingest-semi-structured-data-faster-and-more-efficiently-variant-now-generally-available > This launch includes Variant Shredding, also Generally Available, a performance optimization that uses Predictive Optimization to automatically improve query performance on Variant data. - (single_source) Databricks launched Lakehouse//RT, a new real-time data warehouse designed for operational analytics, BI and app serving, and observability workloads. - supports https://databricks.com/blog/introducing-lakehousert-real-time-performance-unified-lakehouse > We're introducing Lakehouse//RT, Databricks' new real-time data warehouse designed for operational analytics, BI and app serving, and observability workloads. ## Timeline - 2026-08-25T11:03:50.000Z: Kimi K3 from Moonshot AI is now available on Databricks via Unity AI Gateway. Kimi K3 is initially hosted by Databricks in the US, with more regions planned. Databricks offers native access to proprietary models from Anthropic, OpenAI, and Google Gemini alongside Kimi K3. Models available through Unity AI Gateway are fully governed within the secure Databricks perimeter. Kimi K3 is described as the most advanced open-weight model. Databricks completed the acquisition of Panther, an AI SOC platform for modern security operations. Cybersecurity has fundamentally transformed into a data management and AI problem. Attackers are leveraging automation and AI to move faster and launch multi-stage attacks across cloud, identity, and SaaS environments. Variant is now Generally Available in Databricks for faster semi-structured data ingestion. Variant Shredding is GA as a performance optimization using Predictive Optimization for Variant data. Variant seamlessly handles schema changes so teams can ingest data without updating pipelines. Variant is broadly integrated into the Databricks platform for AI workloads including AI Functions.