provider:braket-rigetti-cepheus

AWS Braket — Rigetti Cepheus-1-108Q

Rigetti Cepheus-1-108Q (12 × 9-qubit chiplets, Ankaa-class) · superconducting · 108 physical qubits · 99.10% 2Q fidelity

verified primary

Pricing model

$0.30 per task + $0.000425 per shot. Two orders of magnitude cheaper than trapped-ion competitors.

1.Representative workload costs

Concrete circuit costs for common workload sizes. Where the pricing model is non-linear (per-program minimums, per-month subscriptions), the table surfaces the floor and break-even points.

Table 1 — Workload cost reference

Workload	Cost (USD)	Note
100-shot pilot circuit	$0.34	Effectively the task fee. Per-shot is rounding error.
10,000-shot characterisation	$4.55	Two orders of magnitude cheaper than IonQ Aria for the same shot count.
VQE 200k shots	$85.30	$6K on IonQ, $85 on Rigetti. The fidelity-vs-cost trade-off is decisive at this scale.

2.Structural hardware advantage

108-qubit superconducting via 12 × 9-qubit chiplet architecture. Per-shot pricing is unique to Rigetti in 2026 and reflects production economics rather than spot-market positioning. GA'd on Braket April 2026.

3.Position in the 2026 stack

Production, new generation. Rigetti's chiplet-based scaling is the structural bet for the 100+ qubit superconducting era — Cepheus-1-108Q is the first product to reach GA on this architecture.

4.Best-fit workloads

High-shot-budget workloads (variational, sampling-heavy)
Sampling for noise-resilient algorithms (QAOA, certain QML routines)
Hardware-aware compilation experiments

5.Pricing trap

Watch for: 99.1% 2Q fidelity is one tick below trapped-ion competitors. Deeper circuits need 5-10x more shots to converge. For deep-circuit chemistry workloads, the cost advantage erodes — Aria / Forte may produce useful results at fewer total shots.

Verification note. GA on Braket April 2026. Reserved $4,100/hr.

6.Source

https://aws.amazon.com/braket/pricing/ · verified 2026-06-03