MX3+
Make room for more AI. Not more hardware.
Larger CNNs, transformers, hybrid architectures, smaller language models, and more sophisticated AI pipelines.
MX3+ extends the MemryX platform to support these increasingly demanding workloads, delivering greater capability while preserving the compact, efficient deployment profile that defines MemryX.
Your next product shouldn't need a bigger power budget.
Edge AI is becoming more capable, but the physical constraints of embedded systems are not going away. Power, thermals, board space, and system cost still matter.
MX3+ is being developed to deliver substantially greater AI capability within the same power and silicon footprint as MX3.
That creates room for larger models, transformer-based architectures, richer AI pipelines, and more capable applications without forcing the rest of the system to grow around the accelerator.
MORE FROM THE EDGE
Expand what your product can do.
Substantially greater compute and on-chip model capacity create more headroom for richer AI experiences, more complex processing, and higher capability within the same class of constrained systems.
10x
More compute
10x
More model weight capacity
Same
Power envelope
Relative to MX3
MORE CAPABILITY. SAME CONSTRAINTS.
Built for products that need to do more.
Larger models, richer pipelines, and more advanced AI experiences across the same kinds of products and systems MemryX powers today.
Video Management Systems
Industrial & Robotics
Transportation
Smart Devices
Put more AI into your next product.
MX3+ is targeted for availability in 2027. Talk to our team about future product requirements, roadmap availability, evaluation opportunities, and how MX3+ can expand the AI capability of your next design.
