Polars Cloud

Distributed, fast, cost effective.
Data Processing for any scale.

Deploy and scale data processing workloads. Natively built on top of the popular Polars API that data teams know and love.

available now

later

AWS logo On premise logo

GPC logoAzure logo

now

later

Key benefits

Key benefits

Performance

Performance matters
if you pay per second

Eliminate rewrites when your data outgrows a single machine. Polars now scales natively to multi-node clusters, letting you process any dataset without changing a line of code. Faster, more cost-efficient data processing for every team.

Performance benchmarks on PDS-H SF1000

Dataset size: approx. 1TB, Number of queries: 22

Polars Cloud$0*0m

Glue$0*0m

Polars 16 x m6i.2xlarge (128vCPU/512GiB) / Glue 16 x Glue.2x (128vCPU/512GiB)

* Costs based on total query runtime and provider markup.

Run the benchmarks yourself with the Polars Benchmark repository

Features

Built for data teams
that move fast

Cloud

Control

Compute runs in your cloud environment, close to your data. Eliminate transfer costs, reduce latency, and keep control over your data without compromising performance.

Zero setup

Focus on your workloads, not infrastructure. Easily scale from local to distributed workloads. Pay only for compute time used, scale down to zero.

Integration

Connect from notebooks, orchestrators, or cloud functions. Start executing queries at scale from anywhere with just a few lines of code.

Extensible

Extend Polars with your own custom expressions that are fast and easy to use

Pricing

Only pay for query execution

Start immediately with transparent, usage-based pricing and a generous free tier.

Deploy on AWS

Pay only for compute resources you use per second. No upfront costs, no minimums, scales down to zero.

  • Integrated AWS billing
  • Access any compute on-demand
  • Native cloud integration

Compute

$0.05

vCPU/h

On-Prem

Deploy a distributed Polars cluster within your Kubernetes environment.

  • Deploy on your hardware in Kubernetes
  • Fully customizable Helm chart
  • View your query history in the cloud dashboard

Compute

$0.05

vCPU/h

Includes 10.000 CPU hours per month at no cost.One concurrent cluster, up to 1.024 cores and 64 nodes.

Enterprise

Deploy distributed Polars entirely in your infrastructure. Same scalable performance, zero external dependencies.

  • Deploy on your hardware
  • Meets your compliance standards
  • Works in airgapped environments

Resources

Find all information to start
building today

Documentation

Comprehensive guides on using Polars Cloud, deployment options, and tutorials to get you up and running quickly.

Explore documentation

FAQ

Quick answers to the most common questions to get started, features and pricing.

Browse FAQ

Support

Contact our team to get support for technical questions, implementation guidance, or to discuss product improvements.

Contact support

Run your next Polars query
at scale within 10 minutes

01

Same code, local and production

Write and test locally, then deploy the exact same Polars code at scale. No other APIs, no refactoring and no handovers to other teams required.

02

Efficient at every scale

Process more data on single instances before adding distributed complexity. When your workload grows, scale to distributed execution with minimal overhead.

03

Only pay when queries run

No minimum spend, no idle costs, no surprise bills. You only pay for actual compute used to execute your queries in seconds.

Read the original on cloud.pola.rs ↗