OSS TanbouOn this page
Overview
Dask is a flexible parallel computing library for analytics. High-level collections such as Dask Array, DataFrame, and Bag build task graphs, and schedulers execute those graphs lazily on a single machine or a cluster.
Features and best fit
Based on official documentation; not hands-on tested · Content checked:
Key features
Partition Array and DataFrame workloads into lazy task graphs
Dask collections split data into chunks or partitions and represent operations as task graphs rather than immediately executing them. Calling compute() asks a scheduler to evaluate the required graph.
Install only core features or add array, dataframe, distributed, and diagnostics extras
The package offers a lightweight core plus optional dependency groups for array, dataframe, distributed, diagnostics, and complete installations so environments can include only the required functionality.
Best fit
Before adoption
Tune partition sizes and task granularity to control scheduler overhead
Because execution is organized around task graphs and partitions, workloads with very many tiny tasks can spend a larger fraction of time in scheduling overhead. Choose chunk and partition sizes appropriate to the data and operations.
Sources: [2]
Dask 2026.8.0 requires Python 3.10+ and aligns its distributed extra to the same release line
The 2026.8.0 package metadata requires Python >=3.10 and specifies distributed >=2026.8.0,<2026.8.1 for the distributed extra, so cluster environments should keep Dask and Distributed on matching release lines.
Official sources
- [1]Dask 2026.8.0 README(2026-10-03)
- [2]Dask 2026.8.0 ten-minute guide(2026-10-03)
- [3]Dask 2026.8.0 installation guide(2026-10-03)
- [4]Dask 2026.8.0 package metadata(2026-10-03)
- [5]Dask 2026.8.0 release(2026-10-03)
Supplemental curator note
Dask can extend familiar NumPy and pandas-style workflows to partitioned data and larger-than-memory or parallel computation. Scheduler overhead can dominate small workloads, so inspect task graphs and partition sizes before moving to distributed execution.
Try it in 3 steps
- 1
Create an isolated Python virtual environment
Use Python 3.10 or newer as required by Dask 2026.8.0.
python3 -m venv .venv && . .venv/bin/activate - 2
Install Dask Array 2026.8.0
Pin core Dask and the Array dependency set, including NumPy.
python -m pip install "dask[array]==2026.8.0" - 3
Compute a chunked array lazily
Split 100 elements into four chunks, build the task graph, and execute it with compute. Add the matching distributed release line when moving to a cluster.
python -c 'import dask.array as da; x=da.arange(100, chunks=25); print(x.sum().compute())'
Growth
Growth trends · Last 30 days
13,931 Stars
Trend data is still being collected.
Development activity
Last 90 days · weekly
- Commits (last 30 days)
- 1
- Open PRs
- 307
Development activity is still being collected.
Built with
Categories and tags
GitHub data
GitHub dataView detailed GitHub data
GitHub Topics
- dask
- python
- pydata
- numpy
- pandas
- scikit-learn
- scipy
- Stars
- 13,931
- Forks
- 1,971
- Watchers
- 203
- Open issues
- 1,043
- Contributors
- 416
- Owner type
- Organization
- Primary language
- Python
- License
- BSD-3-Clause
- Repository last updated
- Sep 29, 2026
Related information
Write a related articleShare a guide or use case for this OSS in Markdown. Articles are published after administrator approval.
Explore next
- scikit-learn67,463 Stars
1 shared tag(s) · 1 shared category(s) · Same language
a Python machine-learning library for composing preprocessing, training, model selection, and evaluation through estimators and pipelines
Python - Apache Airflow47,037 Stars
1 shared tag(s) · 1 shared category(s) · Same language
orchestrate data, ML, and AI workflows with code-defined DAGs
Python - pandas49,908 Stars
4 shared tag(s) · 2 shared category(s) · Same language
a Python data-analysis library for reshaping, joining, aggregating, and reading/writing labeled tabular and time-series data with Series and DataFrame
Python - Apache Spark44,112 Stars
4 shared tag(s) · 2 shared category(s)
Process SQL, DataFrames, streams, and machine learning on one distributed engine
Scala - Apache Arrow17,172 Stars
2 shared tag(s) · 3 shared category(s)
share columnar data through a common in-memory format to reduce conversion and copying across analytics systems
C++ - Matplotlib23,319 Stars
2 shared tag(s) · 2 shared category(s) · Same language
Build static, animated, and interactive Python visualizations by composing plots, figures, axes, and rendering backends
Python
Report incorrect information
Tell us if any listing information is incorrect or outdated.