Skip to product information
Beelink GTi14 Ultra AI PC Intel® Core™ Ultra 9 185H (OpenClaw Pre-Installed)
Sale price
$1,099.00
Regular price $1,399.00
The Beelink GTi14 Ultra is a mini PC built around the 16-core Intel® Core™ Ultra 9 185H, with up to 34 TOPS of combined CPU, GPU and NPU AI compute and a PCIe x8 slot for a full external desktop GPU. OpenClaw comes pre-installed, so an AI coding agent is ready the moment you power it on. It suits developers, creators and power users who want one small machine for office work, local AI experiments and, later, serious graphics.
Who it's for
- Developers who want an AI coding agent ready from first boot, running OpenClaw's cloud-connected agent workflows
- Home lab builders who need dual 10Gbps Ethernet for fast file transfers or link aggregation
- Buyers who want a quiet daily PC now and a gaming or GPU-compute rig later, by adding a discrete card through the PCIe x8 slot
- Anyone compiling code, running VMs and multitasking heavily within a 65W TDP envelope
Key features
- Intel® Core™ Ultra 9 185H: 16 cores and 22 threads. Meteor Lake's hybrid design combines performance and efficiency cores for up to 5.1GHz peak turbo and 24MB of cache, all inside a 65W TDP envelope.
- Intel® Arc™ Graphics: The integrated GPU handles hardware-accelerated ray tracing and modern media encode and decode, so light gaming and GPU-accelerated creative work are possible before you add a discrete card.
- Up to 34 TOPS of combined AI compute: The CPU, GPU and a dedicated NPU work together, enough to power Windows Copilot+ features and run lightweight local models alongside OpenClaw's cloud-connected agent workflows.
- PCIe x8 external GPU support: A dedicated PCIe x8 slot connects a full discrete desktop graphics card to this 1.4-liter-class machine.
- Up to 96GB DDR5 and 8TB NVMe: Dual-channel DDR5 5600MHz supports up to 96GB, and dual M.2 2280 PCIe 4.0 slots support up to 8TB combined, so large local models, VM images and project libraries have room.
- Dual 10Gbps Ethernet: Two 10GbE ports give you serious wired bandwidth for file transfers, home lab networking or link aggregation.
- AI-powered 360° voice pickup: A built-in microphone array uses an AI chip to isolate voice commands from background noise across a 5-meter radius, which suits hands-free video calls. Dual DSP-tuned speakers are built in.
- Fingerprint security: A fingerprint sensor on the power button enables fast, secure Windows Hello login and file encryption.
- 32dB near-silent cooling: Bottom-intake airflow keeps the 65W chip cool at very low fan speeds. There are no vents on the top or sides, so the chassis stays clean-looking and quiet under load.
- OpenClaw pre-installed: You can run an AI coding assistant against cloud models immediately, with no separate install.
Specifications
| Spec | Detail |
|---|---|
| Processor | Intel® Core™ Ultra 9 185H, 16C/22T, up to 5.1GHz, 24MB cache, 65W TDP |
| Graphics | Intel® Arc™ Graphics with hardware ray tracing |
| AI performance | Up to 34 TOPS combined CPU+GPU+NPU |
| Memory | Dual-channel DDR5 5600MHz, up to 96GB |
| Storage | Dual M.2 2280 PCIe 4.0 x4 slots, up to 8TB combined |
| Expansion | PCIe x8 slot for external discrete GPU |
| Networking | Dual 10Gbps Ethernet, Wi-Fi, Bluetooth |
| Security | Power-button fingerprint sensor |
| Audio | AI-enhanced 360° 5-meter microphone pickup, dual DSP-tuned speakers |
| Cooling | Bottom-intake airflow design, ~32dB near-silent operation |
| Power | Wide-input 100V–240V power supply |
| Software | OpenClaw pre-installed; Windows 11 Pro |
Good to know
- The NPU suits smaller local inference tasks such as transcription, embeddings and lightweight quantized models, without a separate GPU workstation.
- If maximum local LLM throughput is your priority, AMD's Ryzen AI Max+ machines in our range, such as the GTR9 Pro, push higher raw AI TOPS and unified memory for that exact use case.
Choose the GTi14 Ultra for versatility: strong general CPU performance, a real eGPU upgrade path, dual 10GbE and enough AI acceleration to make OpenClaw useful day to day. If local LLM throughput matters most, look at the GTR9 Pro instead.