Language + Knowledge
Run open language models, create embeddings, search local files, and build retrieval workflows around documents that stay on your computer.
A powerful, useful computer with a point of view, assembled and ready to go out of the box.
Powerful computing internals.
A beautifully designed object.
An operating system you can customize.
What more could you want?
01 / In situComputer 00
A powerful computer for creative work, local AI, and whatever you want to make next.
- IncludedFully assembled computer, EP-2350 FX-MIC, power cable, getting-started guide, and a few bonuses
- AvailabilityShips October 2026
Run models locally.
Generate an image. Transcribe an hour of audio. Ask questions about your own documents. Run an open language model. Experiment with a coding agent.
Do it without sending the work to a cloud model.
Image + Vision
Use Stable Diffusion, Flux, Qwen, and ComfyUI workflows to generate, understand, and work with images locally.
Speech + Agents
Transcribe audio with Whisper, experiment with speech tools, and run coding agents against local projects and files.
385
Everyday models.
Run OpenAI gpt-oss-20b, small and medium language models, SDXL, Whisper, embeddings, local retrieval, and appropriately sized vision models. This is the configuration for most everyday local-AI work.
395+
Memory-hungry models.
Run the same everyday workloads, then add OpenAI gpt-oss-120b, full-precision Flux workflows, Qwen-Image, much larger contexts, and experimental video generation. Its 128 GB unified-memory pool is what lets those larger workloads fit.
Selected results
What local looks like in practice.
Published and community results on Ryzen AI Max systems with the same processors and memory configurations.
| Model / workload | Runs on | Performance | |
|---|---|---|---|
| 385 | 395+ | ||
| OpenAI gpt-oss-20bOpen-weight language model | ✓ | ✓ | ~58 tok/s |
| Stable Diffusion XL1024 × 1024 image | ✓ | ✓ | ~18 sec/image |
| WhisperLocal speech recognition | ✓ | ✓ | Up to ~60× real time |
| OpenAI gpt-oss-120bOpen-weight language model | ✓ | ~38 to 50 tok/s | |
| Flux.1-dev FP161024 × 1024 image | ✓ | ~78 sec/image | |
| Qwen-ImageRoughly 1.5 to 1.8 MP image | ✓ | ~2 to 2.5 min/image | |
Reference results, not guaranteed specifications. Performance changes with model format, quantization, context length, operating system, drivers, software versions, and power settings. We will publish our own tested configurations as we validate them.
Details
01 / Performance
Inside the computer.
| Spec | 385 | 395+ |
|---|---|---|
| Processor | AMD Ryzen AI Max 385 · 8 cores / 16 threads · 3.6 to 5.0 GHz | AMD Ryzen AI Max+ 395 · 16 cores / 32 threads · 3.0 to 5.1 GHz |
| Graphics | AMD Radeon 8050S · 32 compute units · up to 2.8 GHz | AMD Radeon 8060S · 40 compute units · up to 2.9 GHz |
| Memory | 32 GB LPDDR5x-8000 · 256-bit unified memory | 128 GB LPDDR5x-8000 · 256-bit unified memory |
| Storage | 2 TB NVMe SSD | 4 TB NVMe SSD |
| Storage expansion | 1 of 2 M.2 slots filled | 2 of 2 M.2 slots filled |
| Neural processing | Integrated NPU · up to 50 TOPS | |
02 / System
Around the computer.
- Rear I/O
- 2 × USB4-C · 2 × DisplayPort 2.1 · HDMI 2.1
- Networking
- 5Gb Ethernet · Wi-Fi · Bluetooth
- Enclosure
- Teenage Engineering-designed computer-1 aluminum mini-ITX enclosure
- Case material
- 1 mm raw-anodized aluminum
- Case dimensions
- 170 × 190 × 322 mm (W × D × H)
- Operating system
- Omarchy
- Assembly
- Fully assembled and tested by Lab US
- Availability
- Ships October 2026
In the box
Lab Computer 00
Fully assembled, configured, and tested. Create your owner account when you turn it on for the first time.
Teenage Engineering mic
EP-2350 FX-MIC with its curly 3.5 mm cable.
Power cable
The regional cable selected with your order.
Getting-started guide
A concise printed guide to setup, ports, care, and support.
A few extras
Useful surprises from the workshop, still being decided.
Made in the Mojave
Built here. Tested here. Shipped from here.
Every Lab Computer 00 is assembled, configured, and tested in our California workshop before it heads out into the world.
We are building this machine as the foundation for a longer experiment in useful, expressive computers.
Why we're making thisLab Computer 00 is an independent Lab US product. Component names and specifications belong to their respective makers; no partnership or certified system compatibility is implied.
Frequently asked questions.
What is Lab Computer 00?
It is a fully assembled Lab US computer built around your choice of Framework’s Ryzen AI Max 385 or Ryzen AI Max+ 395 mainboard, a Teenage Engineering-designed computer-1 aluminum enclosure, and Omarchy.
When does it ship?
Lab Computer 00 is planned to ship in October 2026.
How will preorders work?
The current plan is a $1,000 deposit, with the remaining configured balance due before shipment. Final cancellation and refund terms will be published before preorders open.
What configurations are available?
The 385 configuration includes 32 GB of memory and one 2 TB NVMe SSD, leaving one M.2 slot open. The 395+ configuration includes 128 GB of memory and two 2 TB NVMe SSDs for 4 TB total, filling both M.2 slots.
What is Omarchy?
Omarchy is an opinionated Linux environment based on Arch Linux, Hyprland, and Quickshell. Lab Computer 00 ships with it preloaded and prepared for secure owner setup. Read more about our relationship.
Can the computer be upgraded?
Heck yeah. The entire motherboard can be removed and replaced. Both configurations have two accessible M.2 slots; the 385 ships with one filled and one open, while the 395+ ships with both filled. The build is planned with room at the rear for additional ports and future hardware. The 32 GB and 128 GB LPDDR5x memory configurations are integrated and cannot be upgraded separately.
What warranty and return policy will apply?
Final warranty, preorder, and return terms will be published before orders open. You can review the current policy outline or contact Lab US with a question.