download mirror

More information on this software may be found at

Showing directory: /
Parent Directory

Prime95 has built-in assignment handling using the PrimeNet v5 API. Users should use the PrimeNet API when it is available for their GIMPS software instead of the legacy manual testing/assignment pages. Using the PrimeNet API provides many advantages for users, including:

An automated assignment handler program/script that uses the PrimeNet API is also available for most other GIMPS primality testing software, including GpuOwl for GPUs and Mlucas for ARM and other non-x86 CPUs. It was written by Teal Dulcet and Daniel Connelly and adapted from an older script for Mlucas by Loïc Le Loarer and Ernst W. Mayer. As with Prime95, this program will automatically get assignments, report assignment progress and results and upload proof files to the PrimeNet server. It supports Windows, macOS and Linux. Below are some basic instructions to download and setup the script:


Windows users on x86 systems can click one of these links to download a standalone executable that packages the latest versions of both Python and the script. Click this link for how to determine if you are running a 32 or 64 bit version of Windows.

Just extract the ZIP file and save the primenet.exe executable to the same directory as your GIMPS software. Then, run: primenet --help to verify that everything works. Lastly, skip down to the Setup section below and be sure to use “primenet” instead of “python3 -OO” in those commands.

All other users should follow these two steps to install Python and download the script:

  1. Install Python: macOS and Linux should already include Python. Windows users can download an installer from
  2. Download the script: Click this link to download the latest version of the script from the mirror. Just save the file to the same directory as your GIMPS software. Then, run: python3 --help to verify that everything works. Windows users may need to use python or py instead of python3.


  1. Register your system with PrimeNet: Run the below command. See the --help output above for the full list of available options and the respective default values. The script defaults to Mlucas, so if using GpuOwl be sure to also add the --gpuowl option. It saves its configuration to a “local.ini” file by default, so it is only necessary to give these options the first time it is run.
    python3 -OO -t 0 -u <PrimeNet User ID> [-H <computer name>] [-T <work preference>] [--num-workers <num>]
    Supported work preferences (click to show) 4 - P-1 factoring
    100 - First time LL tests
    101 - Double-check LL tests
    102 - World record LL tests
    104 - 100 million digit LL tests
    150 - First time PRP tests
    151 - Double-check PRP tests
    152 - World record PRP tests
    153 - 100 million digit PRP tests
    154 - Smallest available first time PRP that needs P-1 factoring
    155 - Double-check using PRP with proof
    160 - First time PRP on Mersenne cofactors
    161 - Double-check PRP on Mersenne cofactors

    For example, for GpuOwl: python3 -OO -t 0 -u ANONYMOUS -H example -T 150 --gpuowl
  2. Start the script: Run the below command each time you want to start the script. Advanced/power users may want to add the -d/--debug option for more verbose output.
    python3 -OO [-t <update frequency (sec)>] [--checkin <checkin frequency (hours)>] [-D <directory>]...
    For example, for GpuOwl (with the default 1 hour update frequency and 6 hour checkin frequency): python3 -OO

For more information about this program/script and for support, please see the dedicated thread on the forum. Note that there are many other older programs/scripts that may be bundled with the download for your GIMPS software and appear similar to this script, but they are just simple wrappers around the legacy manual testing/assignment pages and do not actually use the PrimeNet API. If the script/program requires you to provide your GIMPS password, it is not using the PrimeNet API and therefore does not provide most of the benefits listed above.

File NameFile SizeDate ModifiedHash
Generated 27 Feb 2024 4:56am