One framework for all your data processing

From millions of rows on your laptop to production workloads across a cluster, Polars is the fastest way to work with data.

Get started Talk to us

Trusted by data and AI teams around the world

Solutions

The compute engine that scales with you

Write your query once. Run it on your machine or, unchanged, at scale on a cluster or GPUs.

Polars Open Source

Start here: the fastest way to process data on a single machine

Polars is the fastest open-source compute engine for Python and SQL, free forever under the MIT license. Install it in seconds and transform millions of rows on your own hardware.

Explore Open Source 40k+

Polars Nanook

Scale out: the same code, managed infrastructure

Nanook is the compute platform that speeds up development for your data teams, managing the infrastructure for on-demand, batch and agentic workloads.

Explore Nanook

Code window with a Polars query that scans Iceberg, filters, groups and aggregates, over a teal aurora

Polars Open Source

Start here: the fastest way to process data on a single machine

Polars is the fastest open-source compute engine for Python and SQL, free forever under the MIT license. Install it in seconds and transform millions of rows on your own hardware.

Explore Open Source 40k+

Polars Nanook

Scale out: the same code, managed infrastructure

Nanook is the compute platform that speeds up development for your data teams, managing the infrastructure for on-demand, batch and agentic workloads.

Explore Nanook

Code window where a Polars query runs remotely on a Nanook compute context with 16 GB of memory, 12 CPUs and a 32-node cluster, over a violet aurora

Why Polars

Built from the ground up for performance

Polars
34.63 sec
DuckDB 2.0
42.12 sec
DataFusion
78.04 sec
Pandas
DNF

Compare

The fastest engine on a single machine and a cluster

Polars plans every query and strips out the wasted work for you, so you get answers faster on smaller machines. That means lower bills and less time waiting on results.

See the benchmarks

1import polars as pl
2
3customers = pl.scan_iceberg(...)
4
5query = (
6    pl.scan_iceberg(...)
7    .filter(pl.col("order_date").dt.year() == 2025)
8    .join(customers, on="customer_id")
9    .group_by("region")
10    .agg(
11        pl.col("revenue").sum(),
12        pl.len().alias("orders"),
13    )
14    .sort("revenue", descending=True)
15)

Docs

Write queries your way: SQL or Python

Whether you write SQL or DataFrames, your query runs on the same Polars engine, with the same leading performance. If you know pandas, Spark or SQL, it feels familiar from day one, only faster.

Read the docs

Padlock on a tile wired into a circuit of lines

Compliance

Deploy and run it entirely in your infrastructure

Your data never leaves your environment, so your security and compliance teams stay in control. Backed by enterprise support and SLAs for teams that work at the limits.

Talk to us

Integrations

Fits the stack you already have

Polars integrates with the data ecosystem. It reads and writes every data format and source.

Databases

Cloud storage

Data formats

Catalog

IPC

The framework AI already understands.Polars' structured code is exactly what AI reads and writes best, an edge that grows within your agentic workflows.

Community

Built in the open, with a community to match

Downloads900M+

GitHub stars40k+

Community members5,000+

Antonio C.Staff Data Engineer @ Just Eat Takeaway
“Polars follows most of other DataFrames tool methodologies. Moving from Spark to Polars felt super easy. I would never even try use Pandas again”
Nader K.Director @ NVIDIA
“I'm so excited about Polars”
Rahul S.Data Scientist @ Apple
“I don't even remember the last time I used pandas. Polars ftw.”
Marcus M.Backend Engineer
“I've been using Polars for a while, and it's easily state-of-the-art: fast, lightweight, and super intuitive. Thank you guys, this is huge for the community!”
Antonio C.Staff Data Engineer @ Just Eat Takeaway
“Polars follows most of other DataFrames tool methodologies. Moving from Spark to Polars felt super easy. I would never even try use Pandas again”
Nader K.Director @ NVIDIA
“I'm so excited about Polars”
Rahul S.Data Scientist @ Apple
“I don't even remember the last time I used pandas. Polars ftw.”
Marcus M.Backend Engineer
“I've been using Polars for a while, and it's easily state-of-the-art: fast, lightweight, and super intuitive. Thank you guys, this is huge for the community!”

Alejandro S.AI & Tech Executive @ Zalando
“Congrats to the polars team! We love Polars at Zalando”
Michal M.AI & Data Solution Architect
“Polars is a badass. Started using it a while ago. Never looked back.”
Esmael A.Data Scientist @ RSA
“Love your work guys, Just love it.”
Casey H.Machine Learning Engineer @ G-Research
“The speedup of Polars compared to Pandas is massively noticeable. I generally enjoy writing code that I know is fast.”
Alejandro S.AI & Tech Executive @ Zalando
“Congrats to the polars team! We love Polars at Zalando”
Michal M.AI & Data Solution Architect
“Polars is a badass. Started using it a while ago. Never looked back.”
Esmael A.Data Scientist @ RSA
“Love your work guys, Just love it.”
Casey H.Machine Learning Engineer @ G-Research
“The speedup of Polars compared to Pandas is massively noticeable. I generally enjoy writing code that I know is fast.”