System Requirements

Here's what you need to run Dome Terminal. Honestly.

Minimum specs get you charting, journal, bots, and basic AI features. Recommended specs let you run a local 7B+ AI model alongside active charting without feeling the squeeze. Here's what each spec actually means for your experience.

Minimum

Minimum Configuration

You'll get charting, journal, alerts, bots, and small local AI models. You'll feel the limit if you push into bigger models or run multiple bots simultaneously.

OS Windows 10 64-bit (21H2 or later) Build 19044+ required — most Windows 10 machines already qualify
CPU Intel Core i5 (8th gen) or AMD Ryzen 5 3600 Fine for charts and journal. You'll notice the limit running local AI inference at the same time.
RAM 8 GB Enough for charting, journal, and basic AI features. Add a local AI model and you'll feel the constraint.
Storage 15 GB free SSD space SSD is required — HDD will make charts and journal painfully slow to load.
GPU Integrated graphics (Intel UHD / AMD Radeon Vega) Enough for the terminal itself. Local AI inference will use CPU only — noticeably slower for larger models.
Internet Broadband connection (5+ Mbps) Needed for live prices, remote AI, and news. Everything else works offline.
Display 1920 × 1080 (Full HD) Lower resolutions work but you'll run out of screen space for panels quickly.
Audio Standard audio output Required for Smart Alert sound notifications — useful when you're away from the screen.
Recommended

Recommended Configuration

Comfortable with a local 7B+ AI model running alongside active charting and a few bots. This is where the terminal stops feeling constrained.

OS Windows 11 Pro (latest) Better memory management and DirectML GPU acceleration for local AI. Worth upgrading if you're doing heavy local model work.
CPU Intel Core i7 (12th gen) or AMD Ryzen 7 5800X 8+ cores means charts, multiple bots, and AI inference can run together without fighting for resources.
RAM 32 GB DDR4 (3200 MHz+) Comfortable with a 7B+ local model running alongside active charting and Python Lab. A 7B model alone uses ~6–8 GB.
Storage 100+ GB NVMe SSD Each 7B model is ~4 GB. You'll want space for a few models, extended candle history, and research data.
GPU NVIDIA RTX 3060+ or AMD RX 6700+ (8 GB VRAM) A dedicated GPU isn't required for the terminal itself. It becomes useful for local AI — 7B models run 5–10× faster with GPU acceleration vs. CPU only.
Internet 100+ Mbps fiber connection Low-latency matters for orderflow data. Fast connection also helps when downloading new AI models.
Display 1440p or 4K, dual monitor Second monitor for journal and alerts while your primary shows charts — you'll use both constantly once you're in the habit.
UPS Uninterruptible power supply recommended If you're running live bots, a power cut mid-session is a real risk. A UPS is cheap insurance.
Feature Compatibility

What minimum actually means in practice

Not all features need the same hardware. This table shows what "minimum" and "recommended" mean for each feature — not just whether it runs, but how it feels.

Feature Minimum Recommended Notes
Core Trading ✓ Full ✓ Full Charts, journal, alerts, orderflow, session management — all run fine on minimum specs
Quant Brain ✓ Full ✓ Full Runs entirely locally, no GPU required. Fast on any supported hardware.
Market Intelligence ✓ Full ✓ Full Requires internet for live news and calendar feeds
Local AI — 0.5B models ✓ Fast on CPU ✓ Full Quick responses on any hardware above minimum. Good for rapid market questions.
Local AI — 1.5B models ~ Noticeable wait ✓ Full Technically possible on minimum specs, but response times are slow enough to interrupt your flow.
Local AI — 7B+ models ✗ Not practical ✓ GPU accelerated Technically possible on minimum specs, but slow enough that you'll find yourself waiting. Needs a GPU with 8+ GB VRAM for usable speed.
Python Lab ✓ Simple notebooks ✓ Full Simple notebooks run fine on minimum. Heavy walk-forward runs on large datasets need 16+ GB RAM.
Dome Script Editor ✓ Full ✓ Full Lightweight — no GPU needed, runs on any supported hardware
Bot Manager — 1–2 bots ✓ Full ✓ Full Low CPU load per bot at normal update intervals. Fine on minimum specs.
Bot Manager — 5+ bots ~ Stick to 1–2 ✓ 5+ bots CPU usage scales with bot count. On minimum specs, 5+ bots will compete with charting and AI for resources.
Walk-Forward Validation ✓ Small datasets ✓ Full Short backtests run fine. Multi-year tick data validation needs 16+ GB RAM to stay comfortable.
4K Multi-Monitor ✗ Not supported ✓ Full Needs a dedicated GPU for smooth 4K rendering — integrated graphics can't sustain it reliably
ORACLE Committee (cloud) ✓ Full ✓ Full Runs on Anthropic's infrastructure — internet required. Hardware doesn't affect response quality.
Network

What needs internet, what works offline

You need internet for live prices, remote AI, and news feeds. Everything else — charts from cache, journal, script editor, local AI, and risk tracking — works offline. Here's the breakdown.

Requires Internet

  • Live prices and orderflow data from exchanges — the core of active trading
  • Market Intelligence news and on-chain feeds from external providers
  • Economic calendar live updates and event countdown sync
  • ORACLE Committee AI — runs on Anthropic's servers, not your machine
  • Software updates and AI model downloads from the model hub
  • Account sync and license validation
  • Webhook-based Smart Alert delivery to external notification services

Works Fully Offline

  • Local AI inference — downloaded GGUF models run entirely on your device, no internet needed
  • Python Lab research using cached or previously imported data
  • Dome Script Editor and script validation — write and test scripts without a connection
  • Trade Journal — all entries are stored locally, always accessible
  • Academy lessons after first download — available offline for review and study
  • Chart replay using cached historical OHLCV data
  • Risk gate logic and Session Management calculations
  • Desktop and audio Smart Alert delivery

Bandwidth usage note

Typical bandwidth during active trading is 50–200 MB/hour, depending on watchlist size and whether orderflow depth data is enabled. Monitoring Level 2 depth across multiple instruments simultaneously pushes that number significantly higher. A 100+ Mbps connection is recommended if you're running multi-instrument orderflow — anything slower and you'll notice latency in the data.

OS Compatibility & Roadmap

Dome Terminal currently runs on Windows 10 and Windows 11 (64-bit only). The application is built on a cross-platform foundation, and native Linux and macOS support is on the roadmap. Linux (Ubuntu 22.04+) compatibility is in internal testing. macOS support is planned for a future major release. All installation files, model downloads, and configuration paths are documented for Windows in the Installation Guide. Running Dome Terminal on Windows via WSL2 is not officially supported.

Ready to download. Start here.

Download Dome Terminal and follow the Installation Guide to get your first chart up in minutes. Tutorial 01 takes 30 minutes and walks you through a complete session — chart to journal to review.

Download Terminal Installation Guide