System requirements
What you need to run Doxwell
Doxwell v1.0 runs on Linux x86_64. Below are the minimum specs to boot and the recommended specs for comfortable day-to-day use. Your hardware is yours — Doxwell does not phone home with documents.
Operating system
Minimum
Linux, kernel 5.10 or newer
Recommended
Linux, kernel 6.0 or newer (Ubuntu 24.04, Fedora 40, Debian 13)
CPU
Minimum
x86_64 Intel or AMD, post-2018
Recommended
8 or more cores, x86_64 with AVX2
RAM
Minimum
8 GB
Recommended
16 GB or more
Free disk space
Minimum
5 GB
Recommended
20 GB SSD (grows with your document library)
GPU
Minimum
Not required
Recommended
Optional — CUDA GPU with 8 GB VRAM or more for local LLM acceleration
LLM provider
Minimum
API key (Anthropic Claude, Mistral, or OpenAI) OR a local Ollama instance
Recommended
Anthropic Claude API or a GPU-backed local Ollama with a capable model
Network
Minimum
Outbound HTTPS to your LLM provider and to doxwell.ai for telemetry. No inbound ports required.
Recommended
Stable broadband; air-gapped operation is possible with a local Ollama and telemetry disabled.
| Minimum | Recommended | |
|---|---|---|
| Operating system | Linux, kernel 5.10 or newer | Linux, kernel 6.0 or newer (Ubuntu 24.04, Fedora 40, Debian 13) |
| CPU | x86_64 Intel or AMD, post-2018 | 8 or more cores, x86_64 with AVX2 |
| RAM | 8 GB | 16 GB or more |
| Free disk space | 5 GB | 20 GB SSD (grows with your document library) |
| GPU | Not required | Optional — CUDA GPU with 8 GB VRAM or more for local LLM acceleration |
| LLM provider | API key (Anthropic Claude, Mistral, or OpenAI) OR a local Ollama instance | Anthropic Claude API or a GPU-backed local Ollama with a capable model |
| Network | Outbound HTTPS to your LLM provider and to doxwell.ai for telemetry. No inbound ports required. | Stable broadband; air-gapped operation is possible with a local Ollama and telemetry disabled. |
Platforms coming later
Signed macOS and Windows packages are in preparation. Until then, Doxwell on those platforms requires a manual build from source (macOS) or WSL2 + the Linux AppImage (Windows).