Pricing

Start at zero.
Scale by evidence.

Simulation is free within the Free limits: up to 100 simulation runs a month, one at a time, up to 60 seconds each, on CPU only. Beyond them, or on paid hardware, you pay in credits — estimated first, reserved before it spends, settled for what it used.

SynapseX Chat writing a typed Python function with doctests
SynapseX Chat, recorded
Simulators
Free for 100 runs a month, then credits
Out of credits
An honest HTTP 402, with a top-up link
Top up
1 credit = $0.01, pay for what you use
Paid runs
Estimated first, settled for what the run used
  • Free

    Explore

    $0forever

    For every scientist, student and tinkerer.

    • Labs, Chat and agents
    • Quantum simulators on CPU, 100 runs a month
    • Sealed evidence on every run
    • Community support
  • Pay-as-you-go · 1 cr = $0.01Most popular

    Researcher

    Creditstop up any time

    For researchers and startups running experiments.

    • Everything in Explore
    • Partner QPU runs, billed per shotNext
    • CPU jobs
    • Budget envelopes with hard stops
    • Priority queues
  • Talk to us

    Team / Enterprise

    Customtalk to us

    For labs, enterprises and platform builders.

    • API keys and entitlements
    • Tenant isolation
    • Audit-grade ledger exports
    • Private backends
    • Priority email support
    • Procurement-friendly invoicing

Failed runs release their reservation automatically — it's in the ledger, not in a promise. The full, append-only ledger is queryable via API.

GPU & VM runs — join the waitlist. No SKU appears here before the hardware behind it does.

Status, honestly

What's live.
What's next.

Compute

The laboratory floor
  • Free simulator probesCPU · 0 credits · exact to 24 qubits
  • Partner QPUsper shot · via gateway providersNext
  • GPU simulationGPU-backed simulator runsNext
  • Managed HPC executionEuropean supercomputersNext

Engine & trust

SoftQuantus
  • QCOShybrid scientific engine · SDK · CLI · API · MCP
  • Sealed evidenceSHA-256 Merkle-sealed run records
  • Platform APIplatform.synapsex.ai/api/v1 · credits & API keys

Products

SynapseX
  • SynapseX Labsresearch workspace · chat.synapsex.ai/workbench
  • SynapseX Chatchat.synapsex.ai
  • SynapseX Console & Code CLImacOS · Windows · Linux · one-line install
  • SynapseX-Q1 modelsresearch modelsEarly access

No vaporware in the pitch: every claim on this site maps to a layer you can see the state of. Simulation and evidence are live today — the rest is on our roadmap.

FAQ

Frequently asked questions
about the SynapseX ecosystem.

SynapseX is a scientific ecosystem for research, experiments and simulation: simulators, modelling tools, a quantum lab, and QCOS, our hybrid, hardware-agnostic scientific engine, offered as an SDK, CLI, API and an MCP server you run yourself. Chat, agents and our models are products inside it, not the whole of it. Simulators and sealed evidence are live today; partner QPU access, GPU and HPC compute are in development.

No. Describe the problem in your domain's language. QCOS checks whether quantum is a sensible candidate before you spend anything, and often the honest answer is to use classical methods. That result is a suitability check, not a speed-up claim.

The Free plan has no time limit. It includes quantum simulators with sealed evidence: up to 100 simulation runs a month, one at a time, up to 60 seconds each, on CPU only. That's the product, not a trial. Past those limits, or on paid compute, you pay in credits.

Estimate → your approval → atomic reservation → execution → settlement for actual usage. Over-reservation is released, never silently kept. The append-only ledger is queryable via API.

A record of the run — backend, seed, parameters, raw results, cost — sealed under a SHA-256 Merkle root. Recomputing the root detects any alteration, down to a single byte, so the record is tamper-evident and lets anyone with access check it was not changed. The seal shows integrity, not who produced the record.

Yes. QCOS is offered as an SDK (softqcos-sdk), a CLI (softqcos), a REST API and an MCP server (softqcos-mcp) that you run yourself. Any MCP-capable agent can list backends, estimate, submit jobs and fetch evidence. Our models are one way into the lab, not the only one.

Today QCOS runs on CPU simulators: exact state-vector simulation up to 24 qubits. Partner QPU gateways, GPU simulation and managed HPC execution are in development. The catalog only lists what is live.

Tenant isolation end to end — your jobs, evidence and ledger are scoped to your organization. We do not train our models on your private research without an explicit, revocable opt-in; the full data-use policy is published before general availability.

Running jobs finish on their reservation. New paid jobs return an honest HTTP 402 with a top-up link. Nothing ever runs unfunded.