Start at $19/mo with CPU solvers. Scale to A100 GPUs and quantum circuits on demand. Pay only for what runs.
Developer
CPU onlyFor solo builders
- CPU solvers (annealing, hybrid, tabu)
- Up to 200 variables per problem
- 100 jobs / month
- REST API access
- Job history (30 days)
- GPU acceleration (A10G)
- QAOA / VQE quantum solvers
- WebSocket progress stream
- Team seats
- Priority queue
- SLA guarantee
Pro
GPU A10GFor teams shipping to production
300s GPU compute included · then $0.008 / GPU-second
- All CPU solvers
- Up to 2,000 variables per problem
- Unlimited jobs
- REST API + WebSocket streams
- Job history (90 days)
- GPU acceleration — 300 GPU-sec / month
- QAOA / VQE quantum solvers
- 3 team seats
- Priority queue
- SLA guarantee
Business
GPU A100 80GBFor serious optimization workloads
2000s GPU compute included · then $0.05 / GPU-second (A100)
- All CPU solvers
- Up to 20,000 variables
- Unlimited jobs
- REST API + WebSocket streams
- Unlimited job history
- A100 80GB — 2,000 GPU-sec / month
- QAOA / VQE quantum solvers
- 10 team seats
- Priority queue
- Uptime SLA
Enterprise
On-premise deployment, custom SLAs, unlimited scale, SSO, RBAC, audit logs, and a dedicated solutions engineer.
Full comparison
| Feature | Developer | Pro | Business |
|---|---|---|---|
| Compute | |||
| CPU solvers | |||
| GPU A10G acceleration | |||
| GPU A100 80GB | |||
| QAOA / VQE (CUDA-Q) | |||
| Included GPU-seconds / month | — | 300s | 2,000s |
| Max problem size (variables) | 200 | 2,000 | 20,000 |
| Usage | |||
| Jobs per month | 100 | ∞ | ∞ |
| Concurrent jobs | 2 | 5 | 20 |
| Max job timeout | 60s | 5 min | 10 min |
| Job history retention | 30d | 90d | ∞ |
| Platform | |||
| REST API access | |||
| WebSocket progress stream | |||
| Python SDK | |||
| Simulation Sandbox | |||
| Benchmark Hub | |||
| Dataset Hub | |||
| Team & Security | |||
| Team seats | 1 | 3 | 10 |
| API key management | |||
| Audit logs | |||
| SSO / SAML | |||
| SLA guarantee | |||
| Dedicated support | |||
Estimate your bill
Typical workloads cost less than you think
TSP, 50 cities
~18s on A10G
$0.01 / solve
Portfolio, 200 assets
~45s on A10G
$0.04 / solve
CVRP, 100 nodes
~90s on A10G
$0.07 / solve
Pro plan includes 300 GPU-seconds (~7 TSP solves / day) before overage kicks in.
Frequently asked
Each solver job reports an actual runtime_s on completion. We bill only that duration — nothing for queue wait time or container spin-up, which we absorb. GPU-seconds are deducted from your included quota first; overage kicks in only once the quota is exhausted.
Jobs keep running. Overage is metered per GPU-second at your plan rate — $0.008/s on Pro (A10G), $0.05/s on Business (A100). Your dashboard shows live usage and remaining quota. You can set a hard monthly spend cap; once hit, GPU jobs are paused until you raise the cap or the billing period resets.
CPU solvers — Simulated Annealing, Hybrid, Tabu Search, and Parallel Annealing — run on our API server at no GPU cost. The GPU Annealing solver dispatches to an NVIDIA A10G (24 GB) on Modal, Multi-GPU Annealing uses an A100 80 GB, and QAOA / VQE run via NVIDIA CUDA-Q on A10G. The routing is automatic based on your chosen solver and problem size.
Developer at $19/mo ($15/mo on annual billing) is the entry tier — CPU solvers on problems up to 200 variables, 100 jobs per month, REST API access, 30-day job history. Solo-builder scale. Upgrade to Pro when you need GPU acceleration or larger problems.
Yes. Annual billing saves roughly 20% — Developer drops to $15/month, Pro to $63/month ($756 billed once yearly), Business to $319/month ($3,828/year). You can switch between monthly and annual at any time from the billing portal; the remaining credit is prorated.
No — NEROX charges from day one. Every plan starts with an active Stripe subscription. Cancel anytime from the billing portal; unused annual credit is prorated back to you. Promotion codes are accepted at checkout if you have one.