21 quantum benchmarks. Hardware-validated. Sealed.

Every benchmark below was executed on an quantum hardware backend, returned an quantum computing job ID, and is D-KaP sealed with full provenance. Click any row to verify the seal cryptographically.

31
Sealed runs
5
On real IBM Heron hardware
16
cuQuantum GPU (simulation)
10
MPS (approximate)
31/31
Ed25519 signatures verified at build
# Substrate Benchmark Backend Qubits IBM Job ID Measured Shots Wall time
001 IBM Heron · real QPU golden_grover ibm_fez 4q d7o1uga94dqs73evlue0 L1_from_uniform 0.817383 · chi2_uniform 5354.6484 · top 0000 @ 0.29541 4096 — verify ↗
002 IBM Heron · real QPU phi_gadt_heron ibm_boston 4q d7not7q94dqs73evbjm0 L1_from_uniform 1.085449 · chi2_uniform 6797.1797 · top 0010 @ 0.228271 4096 — verify ↗
003 IBM Heron · real QPU phi_gadt ibm_fez 4q d7o1i5ik4prs73dssfug L1_from_uniform 1.040527 · chi2_uniform 6254.2188 · top 0010 @ 0.232422 4096 — verify ↗
004 IBM Heron · real QPU phi_qaoa ibm_fez 4q d7o1v2u2jamc73bp30fg L1_from_uniform 0.26123 · chi2_uniform 439.5 · top 1110 @ 0.100098 4096 — verify ↗
005 IBM Heron · real QPU phi_qsvm ibm_fez 4q d7o1upi94dqs73evluo0 L1_from_uniform 1.56543 · chi2_uniform 23640.3438 · top 0000 @ 0.525146 4096 — verify ↗
006 cuQuantum GPU · simulation cuquantum_phi_ansatz NVIDIA GeForce RTX 5090 Laptop GPU 28q — layers 12 · distinct 4096 · top 0111101110101011001001011001 @ 0.000244 4096 0.73s verify ↗
007 cuQuantum GPU · simulation cuquantum_phi_ansatz NVIDIA GeForce RTX 5090 Laptop GPU 28q — layers 16 · distinct 4096 · top 1100101111011000110011001001 @ 0.000244 4096 0.89s verify ↗
008 cuQuantum GPU · simulation cuquantum_phi_ansatz NVIDIA GeForce RTX 5090 Laptop GPU 28q — layers 4 · distinct 4092 · top 1111011110100010101100000001 @ 0.000488 4096 0.36s verify ↗
009 cuQuantum GPU · simulation cuquantum_phi_ansatz NVIDIA GeForce RTX 5090 Laptop GPU 28q — layers 8 · distinct 4095 · top 0110011010101110111010110010 @ 0.000488 4096 0.57s verify ↗
010 cuQuantum GPU · simulation cuquantum_phi_ansatz NVIDIA GeForce RTX 5090 Laptop GPU 24q — layers 12 · distinct 4092 · top 101000100111011101101001 @ 0.000488 4096 0.14s verify ↗
011 cuQuantum GPU · simulation cuquantum_phi_ansatz NVIDIA GeForce RTX 5090 Laptop GPU 24q — layers 16 · distinct 4096 · top 101011110101000101110001 @ 0.000244 4096 0.19s verify ↗
012 cuQuantum GPU · simulation cuquantum_phi_ansatz NVIDIA GeForce RTX 5090 Laptop GPU 24q — layers 4 · distinct 4044 · top 000111100001010010000100 @ 0.000732 4096 0.07s verify ↗
013 cuQuantum GPU · simulation cuquantum_phi_ansatz NVIDIA GeForce RTX 5090 Laptop GPU 24q — layers 8 · distinct 4089 · top 001111001110101110010100 @ 0.000488 4096 0.10s verify ↗
014 cuQuantum GPU · simulation cuquantum_phi_ansatz NVIDIA GeForce RTX 5090 Laptop GPU 20q — layers 12 · distinct 4059 · top 11001100001100001111 @ 0.000488 4096 0.11s verify ↗
015 cuQuantum GPU · simulation cuquantum_phi_ansatz NVIDIA GeForce RTX 5090 Laptop GPU 20q — layers 16 · distinct 4074 · top 10010010000100110000 @ 0.000488 4096 0.13s verify ↗
016 cuQuantum GPU · simulation cuquantum_phi_ansatz NVIDIA GeForce RTX 5090 Laptop GPU 20q — layers 4 · distinct 3880 · top 00011000010110001011 @ 0.001465 4096 0.04s verify ↗
017 cuQuantum GPU · simulation cuquantum_phi_ansatz NVIDIA GeForce RTX 5090 Laptop GPU 20q — layers 8 · distinct 4019 · top 01100101011100111001 @ 0.000732 4096 0.07s verify ↗
018 cuQuantum GPU · simulation cuquantum_phi_ansatz NVIDIA GeForce RTX 5090 Laptop GPU 16q — layers 12 · distinct 3701 · top 0101010101010110 @ 0.001221 4096 0.08s verify ↗
019 cuQuantum GPU · simulation cuquantum_phi_ansatz NVIDIA GeForce RTX 5090 Laptop GPU 16q — layers 16 · distinct 3752 · top 1110110001010101 @ 0.000977 4096 0.09s verify ↗
020 cuQuantum GPU · simulation cuquantum_phi_ansatz NVIDIA GeForce RTX 5090 Laptop GPU 16q — layers 4 · distinct 2837 · top 0001101100000010 @ 0.003662 4096 0.27s verify ↗
021 cuQuantum GPU · simulation cuquantum_phi_ansatz NVIDIA GeForce RTX 5090 Laptop GPU 16q — layers 8 · distinct 3400 · top 0101001110100110 @ 0.001709 4096 0.07s verify ↗
022 MPS · approximate simulation mps_phi_gadt matrix_product_state 40q — chi 128 · distinct 1024 · top 0.000977 · top10 mass 0.009766 1024 0.07s verify ↗
023 MPS · approximate simulation mps_phi_gadt matrix_product_state 40q — chi 256 · distinct 1024 · top 0.000977 · top10 mass 0.009766 1024 0.08s verify ↗
024 MPS · approximate simulation mps_phi_gadt matrix_product_state 40q — chi 64 · distinct 1024 · top 0.000977 · top10 mass 0.009766 1024 0.08s verify ↗
025 MPS · approximate simulation mps_phi_gadt matrix_product_state 30q — chi 128 · distinct 1024 · top 0.000977 · top10 mass 0.009766 1024 0.07s verify ↗
026 MPS · approximate simulation mps_phi_gadt matrix_product_state 30q — chi 256 · distinct 1024 · top 0.000977 · top10 mass 0.009766 1024 0.06s verify ↗
027 MPS · approximate simulation mps_phi_gadt matrix_product_state 30q — chi 64 · distinct 1024 · top 0.000977 · top10 mass 0.009766 1024 0.05s verify ↗
028 MPS · approximate simulation mps_phi_gadt matrix_product_state 20q — chi 128 · distinct 1024 · top 0.000977 · top10 mass 0.009766 1024 0.04s verify ↗
029 MPS · approximate simulation mps_phi_gadt matrix_product_state 20q — chi 256 · distinct 1022 · top 0.001953 · top10 mass 0.011719 1024 0.04s verify ↗
030 MPS · approximate simulation mps_phi_gadt matrix_product_state 20q — chi 64 · distinct 1023 · top 0.001953 · top10 mass 0.010742 1024 0.06s verify ↗
031 MPS · approximate simulation mps_phi_gadt matrix_product_state 20q — chi 64 · distinct 1021 · top 0.001953 · top10 mass 0.012695 1024 0.05s verify ↗

How to verify any benchmark

1. Click any row to see its seal: the receipt filename, its sha256, the seal hash, and the Ed25519 signature over that hash. Every row on this page is generated from its receipt file at build time — the build refuses any row whose signature does not verify.
2. The signature is Ed25519 over the 32-byte hex-decoded seal hash, against the published key. ML-DSA-87 is carried alongside. These are signed archive receipts; they are not anchored to any L2 chain, and this page no longer claims they are.
3. The 5 hardware rows carry real IBM job ids, and all of them appear in the 1,605-job Heron corpus published at molecule-gen.org/evidence. Job lookup against IBM is instance-scoped, so a stranger gets RuntimeJobNotFound — the corpus is the check that needs no account.
4. The 16 cuQuantum and 10 MPS rows are simulation, carry no job id, and are labelled as such. MPS is approximate by construction.