AI Resources
gputaskman
gputaskman is a terminal tool for seeing which programs are using your GPU. If a local model fills your graphics memory, you can check the model runner alongside other apps and see how much memory each process reports.
Where the runtime exposes enough detail, the tool can also show evidence about model weights and the cache used during a conversation.
What it is
GPU memory by process
The terminal view lists active processes and their available GPU counters. It can save a snapshot to a file so you can compare two points in the same run.
Why it stands out
See measurements and estimates separately
The tool separates measured process memory from figures reported by a model runner and estimates based on a model file. That lets you see whether measured process memory changed by as much as the estimated model cache.
Availability
Open source and ready to inspect
The project is published under MIT. The repository has screenshots and setup notes, and PyPI lists the Python package. Check the project table to see which counters are available for your operating system and GPU.
Email updates
See which AI tools are getting attention
Get occasional, source-linked notes on what they do, who they may suit, and what to check before trying them.
Why it matters
What makes it useful
If a model almost fills your GPU, take one snapshot at your current context limit and another after changing it. Use the same GPU for both snapshots and take each after the model answers a request. Some runners allocate memory only when they need it.
What to know
Where it fits
Use it alongside the local app that runs your model. The project documents Windows, Linux, WSL and Apple Silicon macOS. The counters and model details you get depend on the platform, driver and runtime.
Notable points
What stands out
In the project’s Gemma 3 example, a higher context limit raised the cache estimate by 504 MB. The runner’s measured dedicated GPU memory rose by 83 MB. Those figures describe different things, so check the measured change before using a cache estimate to decide how much space you have left.
Before using
What to review
Use the support notes to check the counters your driver exposes. A missing counter means it was unavailable.
Execution details vary by runtime. A model file listed beside a process shows a connection to that file. It does not establish how much of the file is in GPU memory.
Check the original listing's documentation for current requirements, limitations, licensing and any relevant privacy or security considerations.
Reader fit
Who may find it relevant
This fits someone tuning a local model and deciding what to change first. If you only need to know whether the GPU is busy, its overall activity reading may be enough. Use the process view when you need to see where that activity or memory is coming from.
Editorial note
Why LifeHubber lists it
LifeHubber lists gputaskman because its comparison keeps the device total and the changed process rows in view together. If the GPU gets fuller during a trial, check whether the runner’s row grew too. If another app grew while the runner stayed steady, that gives you a reason to check the other app before switching to a smaller model.
Source links
Source materials
Reader note
About our listings
LifeHubber’s listings help you explore what’s available in AI and are not endorsements. We don’t independently verify every detail or guarantee that each listing is accurate, safe, actively maintained, or right for your needs.
More in Ecosystem
Keep browsing this category
Explore more AI ecosystem resources.
Docling
docling-project/docling
A document conversion toolkit with structured text, tables and layout, native chunking for retrieval, Markdown and JSON exports, and local model configuration.
Transformers.js
huggingface/transformers.js
Hugging Face’s JavaScript library for running compatible AI models in a browser, with pipelines for tasks such as speech recognition, translation and image classification.
Surya
datalab-to/surya
A Datalab document OCR and analysis toolkit for OCR, layout analysis, reading order, table recognition, math-aware output, vLLM or llama.cpp backends, and structured document results.
For project maintainers
Listed here? You can use the badge.
If you maintain a project with a current LifeHubber listing, you may add the optional “Listed on LifeHubber AI Resources” badge to its README, docs, or website. No introduction or permission request is needed.
MORE ON LIFEHUBBER
Use AI on your own terms.
Explore more AI projects, stories, guides, tools and community insights across LifeHubber.