Quantum Backends

QCOS provides unified access to quantum hardware from multiple cloud providers. This guide covers all available backends, their capabilities, and best practices for each.

Backend Overview

Backend	Provider	Qubits	Type	Best For
IBM Brisbane	IBM Quantum	127	Superconducting	Production workloads
IBM Sherbrooke	IBM Quantum	127	Superconducting	High-fidelity circuits
IBM Heron	IBM Quantum	133	Superconducting	Latest technology
IonQ Aria	AWS Braket	25	Trapped Ion	High connectivity
IonQ Forte	AWS Braket	36	Trapped Ion	Complex algorithms
Rigetti Ankaa-3	AWS Braket	84	Superconducting	Fast execution
IQM Garnet	AWS Braket	20	Superconducting	European data residency
QuEra Aquila	AWS Braket	256	Neutral Atom	Analog simulation
Quantinuum H1	Azure Quantum	20	Trapped Ion	High fidelity
Quantinuum H2	Azure Quantum	32	Trapped Ion	Enterprise workloads

---

IBM Quantum Backends

IBM Brisbane

job = client.run(circuit, backend="ibm_brisbane")

Specification	Value
Processor	Eagle r3
Qubits	127
Topology	Heavy-hex
T1 (avg)	264 μs
T2 (avg)	128 μs
CNOT Error	0.8%
Readout Error	1.2%

Best for: General-purpose production workloads, VQE, QAOA.

Pricing: $115.20/minute (QCOS Standard)

---

IBM Sherbrooke

job = client.run(circuit, backend="ibm_sherbrooke")

Specification	Value
Processor	Eagle r3
Qubits	127
Topology	Heavy-hex
T1 (avg)	280 μs
T2 (avg)	135 μs
CNOT Error	0.7%
Readout Error	1.0%

Best for: High-fidelity experiments, research validation.

Pricing: $115.20/minute (QCOS Standard)

---

IBM Heron

job = client.run(circuit, backend="ibm_heron")

Specification	Value
Processor	Heron
Qubits	133
Topology	Heavy-hex
T1 (avg)	300+ μs
T2 (avg)	150+ μs
CZ Error	0.3%
Readout Error	0.8%

Best for: Latest technology, lowest error rates.

Pricing: $115.20/minute (QCOS Standard)

Note: IBM Heron uses native CZ gates instead of CNOT. QCOS optimization automatically transpiles circuits for best performance.

---

AWS Braket Backends

IonQ Aria

job = client.run(circuit, backend="ionq_aria")

Specification	Value
Technology	Trapped Ion
Qubits	25 (algorithmic)
Connectivity	All-to-all
1Q Gate Fidelity	99.94%
2Q Gate Fidelity	99.4%
Readout Fidelity	99.3%

Best for: Circuits requiring high connectivity, quantum chemistry.

Pricing: $0.36/task + $0.036/shot

Example Cost: 1000 shots = $0.36 + $36.00 = $36.36

---

IonQ Forte

job = client.run(circuit, backend="ionq_forte")

Specification	Value
Technology	Trapped Ion
Qubits	36 (algorithmic)
Connectivity	All-to-all
1Q Gate Fidelity	99.97%
2Q Gate Fidelity	99.5%

Best for: Complex algorithms, highest gate fidelity.

Pricing: $0.36/task + $0.096/shot

---

Rigetti Ankaa-3

job = client.run(circuit, backend="rigetti_ankaa")

Specification	Value
Technology	Superconducting
Qubits	84
Connectivity	Square lattice
Median 1Q Fidelity	99.5%
Median 2Q Fidelity	95%

Best for: Fast turnaround, large-scale experiments.

Pricing: $0.36/task + $0.00108/shot

Example Cost: 10,000 shots = $0.36 + $10.80 = $11.16

---

IQM Garnet

job = client.run(circuit, backend="iqm_garnet")

Specification	Value
Technology	Superconducting
Qubits	20
Location	Europe (Finland)
Connectivity	Square lattice

Best for: European data residency requirements, GDPR compliance.

Pricing: $0.36/task + $0.00174/shot

---

QuEra Aquila

job = client.run(circuit, backend="quera_aquila")

Specification	Value
Technology	Neutral Atom
Atoms	Up to 256
Type	Analog quantum simulation

Best for: Optimization problems, quantum simulation, Ising models.

Pricing: $0.36/task + $0.012/shot

Note: QuEra Aquila operates in analog mode. QCOS provides a translation layer for digital circuits where applicable.

---

Azure Quantum Backends

Quantinuum H1

job = client.run(circuit, backend="quantinuum_h1")

Specification	Value
Technology	Trapped Ion
Qubits	20
2Q Gate Fidelity	99.8%
Connectivity	All-to-all

Best for: Enterprise workloads, highest reliability.

Pricing: Azure credits + 20% QCOS fee

---

Quantinuum H2

job = client.run(circuit, backend="quantinuum_h2")

Specification	Value
Technology	Trapped Ion
Qubits	32
2Q Gate Fidelity	99.8%
Features	Error correction ready

Best for: Enterprise production, error-corrected circuits.

Pricing: Azure credits + 20% QCOS fee

---

HPC Simulators

LUMI Simulator

result = client.simulate(
    circuit,
    method="statevector",
    noise_model="ibm_brisbane"
)

Specification	Value
Max Qubits (statevector)	40
Max Qubits (MPS)	100+
GPU Acceleration	Yes (AMD MI250X)
Location	Finland (CSC)

Methods:

• statevector - Exact simulation up to 40 qubits
• density_matrix - Up to 25 qubits with noise
• mps - Tensor network up to 100+ qubits (low entanglement)

Pricing: See Pricing

---

Choosing the Right Backend

Decision Matrix

┌─────────────────────────────────────────────────────────────────┐
│                  Which Backend Should I Use?                    │
├─────────────────────────────────────────────────────────────────┤
│                                                                  │
│  High connectivity needed?                                       │
│    YES → IonQ Aria/Forte or Quantinuum                          │
│    NO  ↓                                                         │
│                                                                  │
│  Need many qubits (>50)?                                        │
│    YES → IBM Brisbane/Sherbrooke or Rigetti Ankaa               │
│    NO  ↓                                                         │
│                                                                  │
│  Lowest error rates critical?                                    │
│    YES → IBM Heron or Quantinuum H2                             │
│    NO  ↓                                                         │
│                                                                  │
│  Fast turnaround needed?                                         │
│    YES → Rigetti Ankaa (short queue)                            │
│    NO  ↓                                                         │
│                                                                  │
│  Cost-sensitive?                                                 │
│    YES → Rigetti ($0.00108/shot)                                │
│    NO  → IBM Brisbane (best overall)                            │
│                                                                  │
└─────────────────────────────────────────────────────────────────┘

Cost Comparison (1000 shots)

Backend	Cost
Rigetti Ankaa	$1.44
IQM Garnet	$2.10
QuEra Aquila	$12.36
IonQ Aria	$36.36
IonQ Forte	$96.36

---

Backend Status

Check backend availability:

# All backends
backends = client.backends()
for b in backends:
    print(f"{b.id}: {b.status} (queue: {b.queue_length})")

# Specific backend
backend = client.backend("ibm_brisbane")
print(f"Status: {backend.status}")
print(f"Next maintenance: {backend.maintenance.next_scheduled}")

Status Values:

Status	Description
online	Available for jobs
busy	Online but high queue
maintenance	Scheduled maintenance
offline	Temporarily unavailable
retired	Permanently unavailable

---

Backend-Specific Optimization

QCOS automatically optimizes circuits for each backend:

# QCOS detects backend and applies appropriate optimization
job = client.run(circuit, backend="ibm_heron")

# Check what optimization was applied
print(f"Original depth: {job.metrics.original_depth}")
print(f"Optimized depth: {job.metrics.optimized_depth}")
print(f"Native gates: {job.metrics.native_gates}")

Backend Optimizations

Backend	Native Gates	Optimizations
IBM	CX, RZ, SX, X	Heavy-hex mapping, dynamic decoupling
IBM Heron	CZ, RZ, SX	CZ-optimized transpilation
IonQ	All-to-all	Gate fusion, native gate set
Rigetti	CZ, XY	Parametric compilation
Quantinuum	ZZ	ZZ-based decomposition

---

Multi-Backend Workflows

Execute the same circuit across multiple backends:

backends = ["ibm_brisbane", "ionq_aria", "rigetti_ankaa"]
jobs = []

for backend in backends:
    job = client.run(circuit, backend=backend, shots=1000)
    jobs.append(job)

# Collect results
for job in jobs:
    result = job.result()
    print(f"{job.backend}: {result.counts}")

---

Next Steps

• 📖 Circuit Optimization - Maximize backend performance
• 💰 Pricing - Detailed cost breakdown
• 🔧 API Reference - Backend API endpoints