What changed
Read about changes to the guides, apps, models, and tools on this site.
About these updates
This page records changes that help you choose an app, set up a model, or understand the tools on this site.
Updates include new guides, corrected recommendations, changes to calculator assumptions, and clearer advice about privacy.
Small spelling and layout corrections may not appear here.
Published August 2026 content and catalog refresh
Published four news explainers, refreshed six runtime pages, and expanded the exact-model catalog from 85 to 90 entries.
Read the news explainers for changes to local and cloud features before choosing an app.
Added the first reference entries
Added the first reference information about local AI apps, models, and hardware.
This entry marks the start of the reference collection.
Added initial runtime and tool reference entries
Added initial records for Ollama, LM Studio, Open WebUI, AnythingLLM, Jan, llama.cpp, and GPT4All.
Compare the apps by the work you want to do, such as chat or document search.
Added initial local model-family entries
Added cautious model-family entries for Llama, Qwen, Mistral, Gemma, Phi, and DeepSeek-style local models.
Choose an exact model release and size before checking whether it fits your computer.
Added conservative hardware profiles
Added RAM/VRAM class profiles for Macs, Windows laptops, desktop GPUs, mini PCs, and a 64GB workstation bucket.
Actual memory needs depend on your model, compression setting, conversation length, and other running apps.
Added comparison, guide, and troubleshooting entries
Added initial entries for choosing apps, following setup guides, and solving common problems.
Use the guide or troubleshooting page for the specific task you need help with.