Quantum Platform Stability
Weekly litmus-test circuits run on each platform. We track consistency over time — not a cross-platform ranking. Each platform is benchmarked only against its own prior runs.
Consistency over time
Within-run consistency score (1 - std dev) per run — higher is more consistent. Faded line and dots are individual runs; bold line and larger dots are the 4-run rolling average. Both share the same colour per platform.
Success probability over time
Faded line and dots are individual runs; bold line and larger dots are the 4-run rolling average. Both share the same colour per platform.
Time to results: wall-clock (includes queue)
Wall-clock duration from first circuit submitted to last result received, including any queue wait time on the device. Log scale — each grid line is 10× the previous.
Platform summary
Cost per benchmark run
10 circuits × 100 shots. Pricing as of April 2026.
AQT pricing converted at EUR/USD ≈ 1.09. IonQ Aria-1 figure is historical ($0.03/shot via Braket). Forte-1 (direct) cost based on observed billing; Forte-1 (Braket) at published Braket rates. IBM Pittsburgh and Marrakesh costs are averages based on observed QPU-second billing at $1.60/QPU-second.
Benchmarks run weekly. Each run samples 10 circuits from a family of 24 (6 circuit depths × 4 input states), 100 shots each.
"Paused" means the platform is either retired (hardware no longer available from the provider) or temporarily suspended due to cost. Historical data remains visible on each platform's page.