Use the App comparison tool to compare two to four apps. Select your operating system and the work you want to do, then compare requirements that could change your choice: a graphical interface, a local API, document chat, several users, or willingness to maintain a Docker installation. You do not need every feature to complete a first local chat.
| What you want to do | What to compare first | Useful next step |
|---|---|---|
| Try a downloaded model on one computer | Operating system, installation steps, model support and chat interface | Read the Ollama and LM Studio comparison. |
| Connect another app to a local model | Local API support and the separate app's connection requirements | Get one model answering in the runner before connecting the second app. |
| Ask questions about your own documents | Text extraction, retrieval, source inspection and the chosen model provider | Follow the local PDF chat guide. |
| Give several people a browser workspace | Deployment, accounts, permissions and responsibility for updates | Compare Open WebUI and AnythingLLM. |
A model runner loads and executes a model. A workspace may connect to that runner while adding conversations, documents or user accounts. Those roles can overlap, so start with the smallest setup that covers your task. Adding a workspace does not remove the need for a working model provider.
Read the filters and comparison table
Filters narrow the app choices shown in the tool. The comparison table continues to include every selected app, including apps that do not match the current filters; a notice identifies those mismatches. Use the filters to narrow candidates, then review the selected apps and the mismatch notice before treating the table as a suitable shortlist.
“Show differences only” hides listed details that are equal across the selected apps. It does not test the apps, compare answer quality or establish that two features behave identically. A document-chat entry, for example, does not show whether your scanned PDF will yield readable text or whether the model will cite the right passage.
For a first comparison, select Ollama and LM Studio, choose your operating system, and decide whether another application needs a local API. Open each installation guide before downloading a model. If you later add a browser workspace, follow the Open WebUI and Ollama setup after the runner can answer a short prompt.
Check model fit and the complete data path
An app match does not establish that a particular model fits your computer. Use the exact model file, compression and intended context in the RAM and VRAM calculator. Dedicated GPU memory and system RAM are separate budgets; on an Apple Silicon Mac, the model shares unified memory with the operating system and other apps.
Privacy also depends on configuration. Check where generation, document extraction, embeddings and any connected tools run. A desktop window can send requests to a remote provider, and a browser can connect to a model on your own computer. Use the local and cloud comparison to work through those choices.
Share the comparison you selected
Changing selections updates the App comparison tool. Use Copy link to create an address containing the selected apps and filters. The link does not contain chats or documents, and opening it uses the site's current app information. If an app or filter can no longer be restored, check the notice and displayed choices before using the result.
Return to the App comparison tool to choose apps, apply filters and review the current comparison table.