crawl4ai icon indicating copy to clipboard operation
crawl4ai copied to clipboard

[Bug]: docker hub images are outdated

Open bastian-worx opened this issue 9 months ago • 2 comments

crawl4ai version

0.5.0

Expected Behavior

When pulling the docker image unclecode/crawl4ai:latest, I expect an up-to-date crawl4ai version (some 0.5.0.post... at the moment).

Also, I expect a self-contained image to have fairly recent dependencies, most importantly, python.

The available image should provide tags that relate to the included versions. Finally, tags such as "basic", "all" and "latest" should point to multiarch manifests rather than to arm architecture images.

Current Behavior

As of today, versions in docker hub, pypi and github do not match.

  • github only knows a single "release", which is 0.5.0.post1
  • pypi has 0.5.0.post8
  • docker hub (probably?) uses v0.2.75

No version tags are available, and the "basic" builds provide "basic", "arm64-basic", and "amd64-basic" tags rather than arm64 plus amd64 images and a multiarch "basic" manifest.

Python 3.10, while not yet being EOL, is not an optimal choice for a new project that I want to create. I would suggest referring to Python 3.13.

linux/arm64 and linux/arm64/v8 are probably redundant.

I am aware of the fact that I can build an up to date docker image myself; however, an upstream image would be welcome.

Is this reproducible?

Yes

Inputs Causing the Bug


Steps to Reproduce


Code snippets


OS

Docker

Python version

3.13

Browser

No response

Browser version

No response

Error logs & Screenshots (if applicable)

No response

bastian-worx avatar Apr 02 '25 10:04 bastian-worx

Hi @bastian-worx Thanks for using Crawl4ai and reporting this. That's correct. The new Docker image is under test, and you can run it by cloning the repository and building the image locally. We are almost done with the testing, and we have made a lot of improvements. Hopefully, this month, we will update the images on the hub and start using versioning for ease of use. Please stay tuned and use them when they're ready.

unclecode avatar Apr 02 '25 12:04 unclecode

The new Docker image is under test..

@unclecode Please also make a minimal (cpu, no LLM, no pdf) image with the smallest dependences set for barebones crawling !

Torch images are huge !

injen-jb avatar Apr 03 '25 13:04 injen-jb

@bastian-worx This is now updated. You can pull the latest version from dockerhub with

docker pull unclecode/crawl4ai:latest
# Or specific version
docker pull unclecode/crawl4ai:0.6.0rc1-r1

aravindkarnam avatar May 07 '25 06:05 aravindkarnam