windows-guide

OS Requirements

• Target OS: Windows 11 (User account of Windows Home).
• Unsupported: Enterprise or Server editions.

Installation

1. Download the installer executable.
2. Run the installer.
3. Select “Current User” only. Do not install for “All Users”.
4. To launch, run the AnythingLLM desktop shortcut.

---

Troubleshooting

Error: “AnythingLLM cannot be closed” during install

Cause 1: Application is running in the background (v1.11.0+)

1. Locate the AnythingLLM icon in the system tray.
2. Right-click and select Quit AnythingLLM.
3. Click Retry in the installer.

Cause 2: Previous installation was configured for “All Users”

1. Right-click the AnythingLLM Desktop icon and select Open File Location.
2. Run Uninstall AnythingLLM.exe (run as Administrator if prompted).
3. Re-run the installer and select Current User only. (Note: This process retains user data and settings).

---

GPU Support CDN Installation Failure

Recommended Alternative: Install Ollama separately and connect AnythingLLM to the local Ollama instance.

Manual Extraction Workaround:

1. Locate ollama_install.log in %APPDATA%\Local\Programs\AnythingLLM and identify the <version> string.
2. Download the required archive files:
   • Base Binaries: https://cdn.anythingllm.com/support/ollama/<version>/bins.7z
   • NVIDIA GPU (CUDA v13): https://cdn.anythingllm.com/support/ollama/<version>/cudav13.7z
   • NVIDIA GPU (CUDA v12): https://cdn.anythingllm.com/support/ollama/<version>/cudav12.7z
   • AMD GPU (ROCm/Vulkan): https://cdn.anythingllm.com/support/ollama/<version>/rocm_vulkan.7z
3. Create the directory: %APPDATA%\Roaming\anythingllm-desktop\storage\engines\ollama\lib\ollama
4. Extract bins.7z directly into the directory created in Step 3.
5. Extract the GPU-specific archive (cudav13.7z, cudav12.7z, or rocm_vulkan.7z) into the same directory.
6. Verify the directory structure:

%APPDATA%\Roaming\anythingllm-desktop\storage\engines\ollama
├── lib
│   └── ollama
│       ├── ggml-base.dll
│       ├── ggml-cpu-**.dll
│       ├── cudav13/ (NVIDIA CUDA v13 DLLs)
│       ├── cudav12/ (NVIDIA CUDA v12 DLLs)
│       ├── rocm/ (AMD ROCm DLLs)
│       └── vulkan/ (AMD Vulkan DLLs)
└── llm.exe

7. Create a file named .ollama-version (no extension) in %APPDATA%\Roaming\anythingllm-desktop\storage\engines\ollama\.
8. Write the exact <version> string (e.g., 0.13.0) inside .ollama-version.
9. Clean up by deleting the downloaded .7z files.
10. Restart AnythingLLM.