Quantum Expectations

Quantum error-correction feasibility: success probability, qubit overhead, records, trends.

CommunityOnlineNo sign-inGlobalFreeRead-only

What it can do

    What data it sees

    Do you need an account

    No: the server works without sign-in

    Quantum error-correction feasibility: success probability, qubit overhead, records, trends.

    How to connect

    How to connect Quantum Expectations to Claude

    The server address to paste into Claude settings:

    https://www.quantum-expectations.com/api/mcp

    1. Open Claude (claude.ai or the desktop app).
    2. Go to Settings → Connectors.
    3. Click Add custom connector.
    4. Paste the server address copied below into Remote MCP server URL and click Add.
    5. In a chat, click + → Connectors and switch the new connector on.

    Custom connectors are available on Free, Pro, Max, Team and Enterprise plans (Free is limited to one). On Team and Enterprise an organization Owner adds the connector first under Organization settings → Connectors.

    Authorization

    After you click Add, a sign-in window for the service opens. Sign in with your own account and approve access. Claude never sees your password.

    How to connect Quantum Expectations to ChatGPT

    The server address to paste into ChatGPT settings:

    https://www.quantum-expectations.com/api/mcp

    1. Open ChatGPT in a browser (chatgpt.com). A Plus, Pro, Business, Enterprise or Edu plan is required.
    2. Turn on developer mode once: Settings → Apps → Advanced settings → Developer mode.
    3. Open Settings → Connectors and click Create.
    4. Fill in the form: Name (anything), Description (one line about what the service does), MCP server URL (copy it below).
    5. Under Authentication choose OAuth if the service requires sign-in, otherwise None. Click Create.
    6. In a new chat open + → Apps/Connectors and enable the connector.

    OpenAI has renamed this section before (Connectors → Apps/Plugins). If the label differs, search settings for "developer mode". On Business/Enterprise workspaces an admin must allow custom connectors first.

    Authorization

    On first use ChatGPT opens the service's sign-in window. Sign in and approve access.

    How to connect Quantum Expectations to Cursor

    The server address to paste into Cursor settings:

    https://www.quantum-expectations.com/api/mcp

    Fastest: click Open in Cursor below and confirm the prompt.

    Manually:

    1. In Cursor open Settings → Cursor Settings → MCP and click Add new global MCP server.
    2. Paste the JSON copied below into ~/.cursor/mcp.json (per project: .cursor/mcp.json in the repo root).
    3. Save the file. The server appears in the MCP list; authorize it there if asked.

    Authorization

    If the service needs sign-in, an authorize button appears next to the server in the MCP list.

    Server tool list (13)

    Raw names from tools/list. Only developers need these.

    compute_expectationGiven a quantum circuit (2-qubit error rate p, qubit count n, depth d, optional connectivity class), compute the effective error rate, success probability, and optional surface-code or qLDPC overhead. Without error correction a 2d-lattice connectivity is charged as a routing multiplier on depth (result.routingOverheadFactor); hardwareId supplies the device's class automatically. The response is self-describing (formulas, assumptions, caveats, glossary, SOTA hardware, historic series with source URLs) so an agent can reason from one call. For the inverse ("what hardware do I need?") use compute_required_error_rate; to rank multiple platforms in one call use compare_hardware_scenarios.
    compute_required_error_rateInverse of compute_expectation. Given a circuit (numQubits, compDepth) and an acceptable effective error rate, return the required per-gate logical error rate (requiredLogicalErrorRatePerGate: a number, or null only when the target is genuinely unreachable; never 0) and, for every EC option (no-EC, surface-code per distance, every qLDPC code), the required physical error rate plus the subset of current SOTA hardware that already qualifies. An option whose inverse lands above the code threshold is capped at min(MAX_P, threshold) and carries a `note` (any sub-threshold p satisfies it); `unreachableReason` is reserved for genuinely unreachable options. Answers "what hardware do I need to run this algorithm?".
    compute_fault_tolerant_resourcesGiven an algorithm stated as (numLogicalQubits, tCount) and a physical error rate (or hardwareId), derive the full surface-code + magic-state-distillation footprint from the general laws of the Litinski lattice-surgery cost model (no per-scenario constants): distillation factory choice, tile layout, required code distance, total physical qubits, and wall-clock time. Results are reported under TWO published logical-error fits (conservative + optimistic) because they disagree by 13-268x (d=7 to d=25) - always state both. Returns an explicit `infeasible` block when no cataloged factory or code distance can satisfy the error budget. Computes numbers only: comparing against classical alternatives and concluding "should this run on a quantum computer" stays with you, the calling agent (state the caveats when you do).
    compare_hardware_scenariosRun the same circuit against multiple current SOTA hardware entries in one call so an agent can rank platforms without N sequential compute_expectation calls. Defaults to every entry in list_current_quantum_computers when hardwareIds is omitted.
    list_current_quantum_computersReturn the representative-entry table of current SOTA quantum computers (id, hardware type, physical qubit count, 2-qubit error rate, connectivity class with a note on the coupling graph, source URL). Same data that powers the website's "Current Quantum Computers" table. For the model's capability prediction per entry in QUOPS units, read hardwareContext.currentQuantumComputers[].modelCapability from compute_expectation; for measured scores, list_quops_scores.
    list_quops_scoresReturn the measured QUOPS capability scores (arXiv:2609.12146): per device, the largest random universal circuit size executed at polarization >= 1/sqrt(e) inside the cone w^2 <= s <= w^3, with width, QUOPS rate, architecture (physical, physical-postselected, logical) and source URL, plus the utility-scale targets in the same unit (RSA-2048 and FeMoco). These are measurements, the yardstick the site's own model is checked against: compare them with modelCapability.quopsEquivalent on the compute_expectation hardware context. Vendor and device names appear here because this table is agent-facing only.
    list_hardware_timingsReturn per-platform gate-cycle timings (2Q gate time, readout time, in SI seconds) plus the testbed they were measured on (`representativeDevice`) and the native 2Q gate name, with source URLs. Joins list_current_quantum_computers via `hardwareType`, but the numbers are BEST-CASE DEMONSTRATIONS from small testbeds, not measurements on the joined devices, which run their gates and array readout orders of magnitude slower. Use for ratio analysis or as optimistic lower-bound inputs to runtime estimates and compute_quantum_volume_rate, and say so when you quote a runtime.
    compute_quantum_volume_rateCompute the Quantum Volume Rate (QV/second): QVR = V_Q / (log2(V_Q) * t_2Q + t_meas). First-order estimate of how fast a device prepares one QV-sized square circuit (one native 2Q gate per QV layer + one end-of-circuit measurement). OVERSTATES achievable rate: real compilation inflates the 2Q-gate count per layer; omits reset/SPAM, mid-circuit measurement, and classical-control latency. For a production throughput metric, see IBM's CLOPS (arXiv:2110.14108).
    fit_historic_seriesFit a log-linear trend (ln(value) = slope * year + intercept) to one historic series — fidelity or qubit-count — for one hardware type. Atomic primitive: compose with list_current_quantum_computers, compute_required_error_rate, or your own modelling to answer "when might hardware reach X?". residualStdDev is the BIASED (maximum-likelihood) RMS — divides by n, not (n - 2); on small series (n ≈ 3–5) inflate by √(n / (n - 2)) before building confidence intervals.
    list_example_algorithmsReturn the curated list of example quantum algorithms with published resource estimates (qubit count, depth/gate count, source paper URL). Useful for comparing what algorithms need vs. what hardware can deliver. Each entry carries a `provenance` field: 'published-circuit' means the figure is reproducible from the source, 'attested-estimate' means the source withholds the circuit and the figure rests on the authors' attestation, with a `provenanceNote` giving the specifics. Carry that caveat whenever you quote an attested figure; do not present it as equivalently sourced.
    list_qldpc_codesReturn the catalog of supported qLDPC codes (id, label, family, n, k, d, circuitLevelDistance, ancilla counts, roundsPerLogicalOp, threshold, fitCoefficients {c0, c1, c2}, logicalErrorExponent [alpha = d_circ/2], source URLs, provenance, caveats). The per-block, per-syndrome-cycle logical error rate is p^alpha * exp(c0 + c1*p + c2*p^2) for p <= threshold (the source paper's own fitting form; c1 = c2 = 0 means a plain power law). `provenance` names the table or section each constant was read from. `caveats` is an array of source-level qualifications on the entry's constants (loose distance bounds, values a source marks as assumed, numbers that differ between sources); empty when the sources carry none. Read it before quoting a code's logical error rate as firm. Use a code's `id` as the `errorCorrectionCode` input to `compute_expectation`.
    get_historic_seriesReturn the full historic time series — either two-qubit gate error rates ("fidelity") or physical qubit counts ("qubit-count") — broken down by hardware type. Each datapoint carries a source URL. Use this to extrapolate trends — "when might hardware reach X?" — or pair with fit_historic_series for a log-linear fit on one hardware type.
    get_agent_briefReturn the plain-text site brief describing scope, assumptions, the honesty clause, and the API contract. Mirrors the /agent.txt document served by the website.