The QuaMM operating system is built into the chip. Solution companies and product companies come for it. With a deep enough team, take the layered base system. To move faster, take the turnkey design.
There is a chip and a reference design. Samples in two weeks.
Give everything an eye that can think
QM102V and QM108H. Bulb cameras, dual-lens cameras, peepholes, smart locks, and frames.
MC6350 and QM102V. Glasses that capture right away, a camera design with one or two lenses, and a learning camera that can recognize, read, and talk.
QM102D, QM102V, and QM108H. Supplied to Niu, Ninebot, and Tailg, in that order. First by share in mid-range and high-end clusters.
QM102D and 6870. Photo frames, whiteboards, and color shelf screens. Shelf screens update over Wi-Fi and rank first on cost. Ad players go up to 8K.
QMF1 at 20 T, QMA1 at 40 T. The industrial display and the AI elevator box use QM108H.
A mass-production edge chip, still uncommon in China. It can sit beside an Android chip, or run a large model on its own. The QMF1 board is on the Lingxi community. The QMA1 board is on the Lingxi community.
The QuaMM operating system is on the chip, already used on more than several million devices. If the team is deep enough, build it yourself. If you want it sooner, take the turnkey design.
QuaMM is an AI-era operating system that carries forward decades of work distilled from Elastos.
If the team is deep enough, take the base system package and build the product your way.
To move faster, take the Turnkey design. A sample can be ready in two weeks. The time goes to the scene and the parts you want to change.
The system has an agent, and it handles image and sound together. Algorithms and multimedia acceleration are on the chip. The runtime sits on QuaMing's own operating system. Multimodal models and agent models can be mixed across the device and the cloud. The same runtime covers vision devices through edge compute, and customer models run here too.
No. QuaMing does not sell finished products to end users. Delivery is chips, hardware and software design, or open modules. The brand, the users, and the data stay with the customer.
The public system name is QuaMM OS. QuaMM is the world's first hybrid multimodal, multi-model agent operating system to reach large-scale mass production on the embedded side, and a base OS that large language models and physical models can share.
In the model runtime on QuaMing's own operating system. World-class companies build the on-device model runtime into the operating system. QuaMing builds its runtime on its own system. Multimodal models and agent models can be mixed across the device and the cloud. The same system runs from low-power vision devices to edge compute. The configuration changes, and so do the model size and the algorithms. Customer models and algorithms run in this runtime too.
There is a chip, a reference design, and the system is already on the chip. Samples in two weeks.
QMF1 is INT8 20 TOPS. QMA1 is INT8 40 TOPS. Vehicle-body control stays on the customer's MCU.
Shanghai HQ is Building 2, Room 304, Zhangjiang Microelectronics Port, 690 Bibo Road. Shenzhen is Tower 5A, Shenzhen Bay Science and Technology Ecological Park, Nanshan. Jiangsu is Room 3073, Building M, 18 Changshan Avenue, Jiangyin. Business email: qmpublic@quaming.com.
Email qmpublic@quaming.com. The materials are free.
QuaMing is a company under Fullhan's control. The R&D center is in Shanghai. The work is chip development, releasing chips, free turnkey designs, and the developer community, in close step with Molchip and Fullhan.
