KDB-X: The Next-Gen of kdb+

End analytics complexity with a unified platform where time-series, vector, and AI converge—and data, models, and logic move as one.

Try KDB-X

Unify Your Data Stack. Empower Everyone.

Built on the industry-leading kdb+, KDB‑X delivers high-performance analytics for time‑series and vector data in a single, developer‑first platform. Convert raw data from open formats like Parquet into real-time action without the complexity of fragmented stacks and leverage new features like MCP server to give everyone the ability to query kdb+ data across your organization.

High-Performance
Data Engine

A single cohesive stack enables seamless integration of time-series and vector data with the award-winning kdb+ at its core. Democratize access to data through q, Python, or SQL across your entire organization with no bottlenecks for real-time responses.

Open & Modern
Architecture

A Developer-First architecture built from the ground up supports open data formats (e.g., Parquet), native integration with Python and SQL, and simplified infrastructure management via modules, drastically reducing development cycles and accelerating time-to-decision.

Speed & Scale
For AI Workflows

With a built-in MCP Server, AI libraries, and GPU acceleration, bring your structured and unstructured data together for real-time, cross-modal analytics—so inference, temporal joins, and vector searches all happen in one environment, without bottlenecks.

Build The Ultimate Real-Time Stack

Single System Dual Modal Database
Eliminate silos, unify time-series and vector data for low-latent, high-performing real-time analytics and AI solutions.

Open Data Format, Cross-Asset Reach
Natively query Parquet, ensuring massive columnar data flows across silos and asset classes with ease.

Answers from all your data
Blend structured, unstructured, and streaming so AI and humans get instant, full-context answers.

Powered by kdb+
Built on the columnar, cache-efficient kdb+, which has delivered low-latent time-series analytics in the most demanding environments for over 30 years.

Modern. Modular. Ready For AI.

Features of KDB-X

Native
Parquet

Work with Parquet as a first-class format in KDB-X, with support for direct querying, partitioning, and object store access.

AI
Libraries

Perform fuzzy, hybrid, and advanced time-series search using built-in algorithms for flexible, high-accuracy data matching.

Module
Management

Language-level integration for modules with enhanced namespacing, dependencies, and packaging to simplify development and encourage reuse.

Expanded
Libraries

Carefully curated community and KX modules for integrations with security protocols, monitoring tools, and third-party interoperability.

GPU
Acceleration

Speed up compute-heavy operations such as as-of joins, matrix multiplication, and vector similarity search. (Spring 2026)

Process & Data
Management

Manage q processes, setup multi-root DBs, and utilize inter-process communication with new system-level enhancements. (Spring 2026)

Agentic
& NLP

Utilize an MCP Server to enable agents and language models to interact with KDB-X via structured prompts, query execution, and schema access.

Developer
Tooling

Debug, profile, and validate q code with integrated tools designed to speed up iteration and improve code quality. (Spring 2026)

See our documentation site for continuous roadmap updates.

Start for free and deploy applications built for performance

Using just 1.5% CPU threads and 8% of System Memory, KDB-X was the fastest across aggregation, filtering, and group-by queries.

Powered by kdb+, designed for the modern AI era, and trusted by Capital Market leaders, start building high-performing, commercially unrestricted apps today.

Try KDB-X

34x

faster for average query speed

1,100x

faster for peak query speed

90%

of test cases, including short-range and multi-year datasets, KDB-X was the leader

View KDB-X Benchmarks

Next-Gen Trading Solutions Are Developed on KDB-X

End-to-end TCA on Parquet

Compute full execution analytics—arrival/implementation shortfall, VWAP/participation slippage, venue/broker/algo-wheel performance—directly on Parquet/Iceberg with native q/SQL (no closed TCA black boxes).
Enrich hot in-memory order/trade tables with SIP/direct TAQ, OPRA options, L2 order-book and futures/FX ticks plus reference data & corporate actions; push down partitions and run governed as-of/point-in-time joins, persisting only what you need.
For buy-side execution teams, sell-side algo desks, and quant researchers: faster best-ex reviews and backtests, lower pipeline/storage duplication, and consistent governance with quicker time-to-insight.

Hybrid Vector + Time-Series Search

Find “looks-like” (embedding similarity) and “behaves-like” (temporal pattern) matches in one place to investigate market microstructure and surveillance events.
Store embeddings alongside market telemetry; shortlist candidates via vector similarity, then confirm with temporal shape matching (e.g., DTW) in the same runtime.
For trade surveillance, market-abuse detection, execution analytics, and quant research: fewer false positives and faster root-cause on anomalies (spoofing/layering-style patterns, regime shifts, liquidity gaps).

Agentic AI on Live Data

Build agents that sense → reason → act on real-time market streams enriched with recent history to automate intraday decisions.
Subscribe to feeds, join context from the lake, run models/policies in q or Python, and trigger actions (route/reweight orders, adjust limits/hedges, escalate alerts)—all in one auditable engine.
For trading assistants, smart order routing, intraday risk/limits, and trade-ops: lower latency, simpler stacks, and clear audit trails.

GPU Accelerated Retrieval

Hit sub-millisecond retrieval and model scoring for surveillance, signal matching, and intraday risk checks.
Keep embeddings/features co-located with tick/OB data and use optional CPU/GPU acceleration for ANN search and inference—no separate vector DB or model service needed.
For market-abuse monitoring, venue/client routing personalization, and real-time risk limits: higher throughput with less infrastructure complexity and cost.

KDB-X: The Developer-First Platform

Get Started Faster Than Ever

Paste your license key into a one-line install script for macOS, Linux, or Windows via WSL, and you’re up with the q REPL immediately—no extra tooling required. Optional Python support is a single pip install, and there are offline installers if you can’t reach the internet.

curl -sLO https://portal.dl.kx.com/assets/raw/kdb-x/install_kdb/~latest~/install_kdb.sh && bash install_kdb.sh --b64lic <Your Key>

q for speed. Python & SQL for ease.

Create columnar tables in q or Python and query them with familiar, parameterized SQL on the same in-memory engine. Change to a single data model with many entry points, so you pick the language and keep the speed without ETL, adapters, or schema translation.

In q: 

q)tab:([] sym: 5?`AAPL`MSFT; qty: 1 + 5?100; prices: 100 * 5?1f) 

q)s)SELECT * FROM tab WHERE sym='AAPL' 
 
======================================================= 

# Or, with Python  
table = kx.Table(  

  data = {'sym': kx.random.random(5, ['AAPL', 'MSFT']),  

          'size': 1+kx.random.random(5, 100), 

          'prices': kx.random.random(5, 100.0)}) 

 ======================================================= 

Query with SQL 

kx.q.sql("SELECT * FROM $1 WHERE sym=$2", table, "AAPL")

Zero-copy queries on open-format data

Treat Parquet as a first-class table: load, read, select. Query lake data in place with q/SQL or Python on the same engine. Open format, zero-copy interoperability across object storage means no ETL, adapters, or conversion, delivering a familiar developer experience and faster time to value from a simpler, integrated stack for real-time analytics and AI.

/ Load the parquet module 

q)([pq]):use`kx.pq 

/ Read the fx_quote parquet file into a table 

q)fx_quote:pq `:fx_quote.parquet 

/ Query the fx_quote table 

q)10#select symbol, ask_price, bid_price, ask_volume, bid_volume from fx_quote

Query across structured and unstructured with AI libs

Built-in AI libraries bring vector and time series similarity search into q/Python on the same in-memory engine, so you pattern match in place (e.g., DTW) with no data movement or separate ML stack.

/load AI Libs module 

q).ai:use`kx.ai 

/Create a time series to search against 

q)ts: neg[5.0] + 1000?10.0; 

/Create a query vector to search against the time series 

q)query_vector:neg[5.0] + 10?10.0; 

/Perform Dynamic Time Warping pattern matching 

q).ai.dtw.filterSearch[ts;query_vector;4;0.3;1.4;::]

Explore Why Capital Markets Choose KDB-X

Developer

Benchmarking KDB-X vs QuestDB, ClickHouse, TimescaleDB and InfluxDB with TSBS

We recently ran a series of benchmarking tests to evaluate how KDB-X performs compared to several open-source databases on standard analytical workloads, including aggregation, filtering, and group-by queries.

By Ferenc Bodon Ph.D • 14 November, 2025

Developer

From ticks to tweets: Combining structured and unstructured financial data with KDB-X

Gain a deeper understanding of market drivers and unlock new business insights by combining structured and unstructured data in KDB-X

By Ryan Siegler • 12 August, 2025

Developer

GPU acceleration in KDB-X: Supercharging as-of joins and sorting

Discover how KDB-X uses GPU acceleration to run as-of joins and sorting up to 10× faster with NVIDIA CUDA and cuDF—cutting end-of-day runtimes dramatically.

By Ryan Siegler • 6 November, 2025

Detect crypto patterns with KDB-X Temporal Similarity Search (TSS)

Developer

Tutorial: Detect crypto patterns with KDB-X TSS

Detect crypto price patterns using KDB-X Temporal Similarity Search (TSS) and uncover market trends in real time.

By Samuel Bruce • 31 July, 2025

Build real-time solutions that handle real-time data

Connect streaming, batch, and historical data in a single engine to ship faster.
Analyze live data at ms latency for real-time monitoring, alerts, and automation
Polyglot by default with q, Python, and SQL in one environment for maximum adoption

To get started, visit our Developer Center to download KDB-X and explore tutorials, docs, and our developer community.

Try KDB-X

Demo the world’s fastest database for vector, time-series, and real-time analytics

Start your journey to becoming an AI-first enterprise with 100x* more performant data and MLOps pipelines.

Process data at unmatched speed and scale
Build high-performance data-driven applications
Turbocharge analytics tools in the cloud, on premise, or at the edge

*Based on time-series queries running in real-world use cases on customer environments.

Book a demo with an expert

"*" indicates required fields

Name

This field is for validation purposes and should be left unchanged.

First Name*

Last Name*

Company*

Job Title*

Business Telephone*

Business Email*

Industry*

Country*

How can KX help you?*

How did you hear about us?*

By submitting this form, you will also receive sales and/or marketing communications on KX products, services, news and events. You can unsubscribe from receiving communications by visiting our Privacy Policy. You can find further information on how we collect and use your personal data in our Privacy Policy.

CAPTCHA

KDB-X: Next-gen kdb+ is here – and it’s built different

Apex innovators: How hedge funds can evolve analytics at speed and scale

Benchmarking KDB-X vs QuestDB, ClickHouse, TimescaleDB and InfluxDB with TSBS

From ticks to tweets: Combining structured and unstructured financial data with KDB-X

KDB-X: The next era of kdb+ for AI-driven markets

Unify Your Data Stack. Empower Everyone.

High-Performance Data Engine

Open & Modern Architecture

Speed & Scale For AI Workflows

Build The Ultimate Real-Time Stack

Modern. Modular. Ready For AI.

Features of KDB-X

Start for free and deploy applications built for performance

Next-Gen Trading Solutions Are Developed on KDB-X

KDB-X: The Developer-First Platform

Get Started Faster Than Ever

Explore Why Capital Markets Choose KDB-X

Benchmarking KDB-X vs QuestDB, ClickHouse, TimescaleDB and InfluxDB with TSBS

From ticks to tweets: Combining structured and unstructured financial data with KDB-X

GPU acceleration in KDB-X: Supercharging as-of joins and sorting

Tutorial: Detect crypto patterns with KDB-X TSS

Build real-time solutions that handle real-time data

Demo the world’s fastest database for vector, time-series, and real-time analytics

Start your journey to becoming an AI-first enterprise with 100x* more performant data and MLOps pipelines.

Book a demo with an expert

High-Performance
Data Engine

Open & Modern
Architecture

Speed & Scale
For AI Workflows