This page records what a cited document supports. It does not report hands-on tests of photonic hardware. Optical networking products are omitted unless a source shows an optical calculation.
Entries checked through 5 October 2026. Plans and announcements are not general availability.
Public hardware access
A visitor can run a supported workload on photonic hardware through a documented self-service route.
No verified entry is listed here.
Restricted evaluation
Hardware access exists, but only for approved evaluators or a closed program.
No verified entry is listed here.
Limited demonstration
A hosted demo runs a fixed example. It does not accept an arbitrary workload.
No verified entry is listed here.
Simulator
Software imitates the programming interface or the operation. It does not provide the photonic chip.
Q.ANT
Native Computing Toolkit CPU backend
The September 2026 SDK release includes a simulator running on a standard CPU. The official repository documents a standalone cpu-backend build. This is software access for developing applications; it does not provide access to a photonic chip.
Workload Development and local testing of native photonic operations through C or Python APIs.
Checked . Documentation reviewed; hardware access not tested.
Announced service
A provider has described future hardware access. An announcement is not evidence that the service is open now.
Q.ANT / IONOS
Announced Native Processing Unit cloud access
Q.ANT's September 2026 announcement describes IONOS cloud hardware access opening in the coming months. The reviewed announcement does not establish a publicly available, self-service hardware service today. Availability, pricing, and access terms require confirmation from the providers.
Workload Planned execution of SDK applications on Q.ANT photonic hardware.
Checked . Documentation reviewed; hardware access not tested.
Historical access
Older documentation describes access. Current availability has not been verified.
Q.ANT
2024 handwritten-digit recognition showcase
The September 2024 release announced a web showcase allowing users to select a digit image for inference on an NPU. It describes a limited demonstration rather than uploading arbitrary workloads. This records the historical announcement; current demonstration access has not been tested.
Workload A fixed MNIST handwritten-digit inference demonstration using matrix-vector multiplication.
Checked . Documentation reviewed; hardware access not tested.
LightOn
LightOn Cloud Optical Processing Unit
LightOn's April 2019 research article described remote OPU prototypes for selected users and cloud access by invitation. That dated documentation supports historical restricted access. It does not establish current availability or discontinuation; a present-day hardware-access route has not been verified.
Workload Optical random-feature transforms for machine-learning research, including sentence embeddings.