Polars

Polars is a fast, multi-threaded DataFrame library written in Rust with a first-class Python API, built and maintained by the open-source Polars community. It ranks #12 in trending tech with roughly 3.8% usage share, and it's climbing fast as data teams look for a modern alternative to pandas. Its appeal is speed and scale: a lazy query engine that optimizes your whole pipeline before running it, so it can chew through datasets far larger than fit comfortably in memory with pandas.

Example

Polars' lazy API (scan_csv instead of read_csv) builds a query plan first, then optimizes and executes it in one pass when you call collect().

import polars as pl

# lazily scan a CSV — nothing is read yet
q = (
    pl.scan_csv("orders.csv")
    .filter(pl.col("total") > 50)
    .group_by("customer")
    .agg(
        pl.col("total").sum().alias("total_spent"),
        pl.len().alias("order_count")
    )
    .sort("total_spent", descending=True)
)

# the optimizer plans the whole pipeline, then runs it here
df = q.collect()
print(df.head(10))

Getting started

Polars installs as a single pip package — no separate compiler or Rust toolchain required, the wheel ships precompiled.

# install with pip
pip install polars

# or with uv
uv add polars