Skip to content
qBraid
001/QBRAID OS · FOR QUANTUM COMPUTING COMPANIES

Run your quantum computer as a service.

qBraid OS is software that quantum hardware providers and system operators use to run their systems and make them available to users. It manages access, scheduling and billing, and runs the processor together with NVIDIA accelerated computing, on premises or in the cloud. Setup and operation are designed to be carried out with AI agents, within permissions you set.

Request early accessWatch the film
AVAILABLE NOW TO EARLY-ACCESS PARTNERS
18HARDWARE PROVIDERS ON QBRAID
35K+DEVELOPERS
1,700+COMPANIES
120+COUNTRIES
002/THE FILM · 69 S

From the lab to its first users.

Connect a system, run it with NVIDIA accelerated computing, and open it to developers. qBraid OS in sixty-nine seconds.

003/AGENT-RUN OPERATIONS

Run by agents. Approved by you.

Setup and day-to-day operation are designed to be carried out with AI agents, within permissions the operator sets, so a person approves the steps that change who can use the system.

USERS

Accounts, organizations and access, managed in one place.

RESERVATIONS

Set aside system time for a team or a customer.

METERING

Usage recorded for every user and every job.

BILLING

Charge for access, or turn billing off for internal use.

qbraid os — agent setupAWAITING APPROVAL

OPERATOR

Connect our QPU to qBraid OS.

Detect hardware1 QPU · control electronics
Configure gatewaygateway online
Link NVIDIA GPUsCUDA-Q ready
Enable scheduling & billingaccounts · queues · metering

APPROVAL REQUIRED · OPERATOR

Opening the system to users changes who can run on it. The agent waits for a person to approve this step.

004/WHY QBRAID OS

The software around the processor.

Before anyone can run on a quantum computer, it needs accounts, queues, billing, a cloud API, GPUs for error correction and drivers for its control electronics. Companies building or operating quantum computers typically develop this themselves.

Figure 1. Today: user accounts, cloud access, GPUs for error correction, job queues, billing and control electronics are separate components joined by glue code around the QPU. With qBraid OS: one integrated system providing accounts and SSO, scheduling and queues, metering and billing, cloud and API access, NVIDIA CUDA-Q on GPUs and HPC, and a control-electronics gateway, used through the qBraid SDK and AI agents. (opens the full-size image in a new tab)

Figure 1. Before and with qBraid OS: separate tools and glue code around the processor, replaced by one integrated system.

TODAY

Each company assembles the stack from separate tools and glue code, and maintains it alongside the hardware.

WITH QBRAID OS

Accounts, scheduling, metering and billing, cloud access, the CUDA-Q platform on NVIDIA GPUs and the control-electronics gateway arrive as one system.

IN OPERATION

The same infrastructure qBraid built and runs for its own agentic platform, which provides access to systems from 18 hardware providers.

005/CAPABILITIES

What qBraid OS does.

Each part replaces software a hardware team would otherwise build and maintain itself. Together they run a quantum system for its users.

01

Agent-run operations.

Setup and operation designed to be carried out with AI agents, within permissions set by the operator.

02

Connect any system.

Connect through the processor's existing interface, or directly to the system's control electronics.

03

Hybrid scheduling.

Schedule quantum processors and NVIDIA accelerated computing together, including systems connected through NVIDIA NVQLink.

04

NVIDIA CUDA-Q platform.

Designed to leverage NVIDIA accelerated computing alongside the processor through CUDA-Q, for quantum error-correction decoding and classical processing during each job.

05

Operate as a service.

User management, reservations, usage metering and billing, for a commercial service or for internal use.

06

Developer access.

The qBraid SDK, qBraid AI agents and NVIDIA CUDA-Q, with the same code developers write for other qBraid hardware.

07

Integration points.

Defined interfaces for compilers, optimizers, error-mitigation tools and resource-estimation services, so each integrates once.

08

Deploy anywhere.

From a hardware company's own lab to a supercomputing center, alongside GPU and HPC systems.

006/ARCHITECTURE

Quantum and NVIDIA accelerated computing as one system.

qBraid OS combines qBraid Runtime and qBraid Gateway around your processor, and runs it with NVIDIA accelerated computing under one scheduler.

Figure 2. qBraid OS architecture: the qBraid SDK and AI agents on top of a three-layer stack (qBraid Runtime for scheduling, metering and access; qBraid Gateway for the control-electronics connection; your QPU), linked to NVIDIA GPUs running CUDA-Q for error-correction decoding and pre- and post-processing, and to HPC and CPU classical compute. Platform services: accounts and SSO, scheduling and queues, metering and billing, cloud and API access. (opens the full-size image in a new tab)

Figure 2. qBraid OS architecture.

SDK · AI AGENTS

How developers and agents reach the system. A program calls device.run(circuit); an agent plans, runs and returns results.

QBRAID RUNTIME

Scheduling, queues, metering and access, from qBraid Runtime.

QBRAID GATEWAY

The connection to the hardware, through its processor interface or its control electronics, from qBraid Gateway.

YOUR QPU

The system you built or operate, in your own lab or at a supercomputing center.

NVIDIA GPUS

Quantum error-correction decoding and classical processing during each job, through the NVIDIA CUDA-Q platform, scheduled together with the processor, including systems connected through NVIDIA NVQLink.

HPC · CPU

Classical compute scheduled together with the quantum processor, on premises or in the cloud.

007/DEVELOPERS

Ready for developers on day one.

Developers reach a system on qBraid OS through the qBraid SDK, qBraid's AI agents and NVIDIA CUDA-Q, with the same code they write for other qBraid hardware.

qbraid sdk · python
from qbraid.runtime import QbraidProviderprovider = QbraidProvider()device = provider.get_device("your-qpu")  # your system's device IDjob = device.run(circuit, shots=1000)print(job.result().data.get_counts())
nvidia cuda-q · python
import cudaqcudaq.set_target("qbraid", machine="your-qpu")@cudaq.kerneldef bell():    q = cudaq.qvector(2)    h(q[0]); x.ctrl(q[0], q[1]); mz(q)print(cudaq.sample(bell, shots_count=1000))
REACH

Offer your system to qBraid's developers.

qBraid is used by more than 35,000 developers from over 1,700+ companies and around 1,200 universities, in more than 120+ countries. A system offered on qBraid appears next to the hardware they already use.

INTEGRATION POINTS

Tools integrate once, not once per system.

Defined interfaces let compilers, optimizers, error-mitigation tools and resource-estimation services connect to qBraid OS a single time.

  • COMPILERS
  • OPTIMIZERS
  • ERROR MITIGATION
  • RESOURCE ESTIMATION
008/DEPLOYMENT

Wherever the system is installed.

qBraid OS can be deployed from a hardware company's own lab to a supercomputing center, where the processor runs alongside GPU and HPC systems on premises.

  • HARDWARE LAB
  • SUPERCOMPUTING CENTER
  • ON PREMISES
  • CLOUD
COMMERCIAL SERVICE

Offer it to customers.

  • Open to developers on qBraid

  • Metering and billing built in

  • SDK, CUDA-Q and AI agents from day one

INTERNAL USE

Run it in your own lab.

  • Private to your researchers

  • The same scheduling, GPUs and agents

  • Billing off

009/EARLY ACCESS

Bring your system to users.

qBraid OS is available now to early-access partners. Tell us about your system and how you plan to operate it.

Request early accesscontact@qbraid.com