Simple, Transparent Pricing
One-time purchase. No subscriptions. No API fees. Run unlimited AI inferences forever.
Ships Q3 2026
Solo
Single Device
Perfect for individual professionals
Devices
1x
RAM
16GB
Compute
100 TOPS
Max Model
32B
- Run models up to 32B parameters
- 16GB unified LPDDR5 memory
- 100 TOPS AI acceleration
- 2TB NVMe storage
- Wi-Fi 6E + 2.5G Ethernet
- OpenAI-compatible API
- Thox OS with OTA updates
- Email support
Best for: Developers, researchers, freelancers
Duo
2-Device Bundle
Save $99
Recommended for professional workloads
Devices
2x
RAM
32GB
Compute
200 TOPS
Max Model
70B
- Run models up to 70B parameters
- 32GB combined unified memory
- 200 TOPS distributed compute
- 4TB combined storage
- MagStack™ auto-clustering
- thox-cluster-nano included
- 1 million token context window
- Priority email support
Best for: Healthcare, legal, small teams
Quad
4-Device Bundle
Save $297
Enterprise-grade for demanding workloads
Devices
4x
RAM
64GB
Compute
400 TOPS
Max Model
100B+
- Run models up to 100B+ parameters
- 64GB combined unified memory
- 400 TOPS distributed compute
- 8TB combined storage
- MagStack™ auto-clustering
- thox-cluster-100b included
- thox-cluster-scout (vision)
- Priority support + onboarding
Best for: Enterprises, hospitals, universities
Enterprise
Custom Configuration
For Fortune 500 and government
Devices
8-20x
RAM
128-320GB
Compute
800-2000 TOPS
Max Model
400B+
- Run frontier 400B+ models
- Up to 320GB unified memory
- Up to 2000 TOPS compute
- thox-cluster-maverick included
- Dedicated account manager
- On-site installation support
- Custom SLA & compliance
- Training & integration services
Best for: Fortune 500, healthcare systems, government
Compare Configurations
Choose the right setup for your workload
| Feature | Solo | Duo | Quad | Enterprise |
|---|---|---|---|---|
| Devices | 1 | 2 | 4 | 8-20 |
| Combined RAM | 16GB | 32GB | 64GB | 128-320GB |
| AI Compute | 100 TOPS | 200 TOPS | 400 TOPS | 800-2000 TOPS |
| Max Model Size | 32B | 70B | 100B+ | 400B+ |
| 1M Token Context | - | |||
| Vision Models | ||||
| MagStack™ Clustering | - | |||
| Priority Support | - | |||
| Dedicated Account Manager | - | - | - |
100% Local
Your data never leaves your device. Complete privacy guaranteed.
Zero Ongoing Costs
No subscriptions, no API fees. One purchase, unlimited use.
Scalable
Start with one device, add more anytime with MagStack™.
Team Ready
Share one cluster across your entire team. No per-seat pricing.
Frequently Asked Questions
What is included with each device?
Every Thox.ai Edge Device includes: the device itself, 65W GaN USB-C power adapter, USB-C cable, Ethernet cable (CAT6), quick start guide, mounting bracket kit, and thermal pad set.
How does MagStack™ clustering work?
Simply stack devices on top of each other. NFC antennas detect proximity and initiate handshake, 8 N52 neodymium magnets self-align with sub-millimeter precision, and a 12-pin pogo connector establishes a 10 Gbps data link plus power passthrough. The cluster forms automatically in ~10 seconds with zero configuration.
Are there any ongoing subscription fees?
No. Once you purchase your Thox.ai device(s), there are no subscription fees, API costs, or usage limits. Run unlimited AI inferences forever with zero ongoing costs.
What happens to my deposit if I cancel?
Your deposit is fully refundable at any time before your device ships. Simply contact support to request a refund.
Can I add more devices later?
Yes! You can add devices to your cluster at any time. Each new device seamlessly integrates with your existing setup via MagStack™ technology.
What compliance certifications do you support?
Thox.ai devices support HIPAA (healthcare), GDPR (EU privacy), SOC2 (security), FERPA (education), and CCPA (California privacy). Since all processing is local, your data never leaves your premises.
Thox.ai™, Thox OS™, and MagStack™ are trademarks of Thox.ai LLC. MagStack magnetic stacking technology is Patent Pending.