Free
Free
Free foreverEvaluate and run locally on your own machine.
- 2 CPUs
- Cypher query engine
- LangChain & LlamaIndex
- Community source
- No GPU · no entropy
Free. Pro.
Pro+. Enterprise.
Same C++ engine throughout. What changes is cores, GPU acceleration, entropy compaction, and whether you run the machine or we do.
Pro
$20 / mo
$240 billed annually (Save 20%)Free edition plus 16 CPUs and 1 CUDA GPU retrieval.
- 16 CPUs
- 1 CUDA GPU retrieval
- Free features included
- Node-locked binary
- Single-IP lock & portable
- Business-hours support
Pro+
$40 / mo
$480 billed annually (Save 20%)Pro capabilities scaled to 32 cores, 2 CUDA GPUs, and entropy worker.
- 32 CPUs
- 2 CUDA GPUs retrieval
- Entropy worker
- Node-locked binary
- Single-IP lock & portable
- Business-hours support
Enterprise
Quote
Custom annual agreementUnlimited cores, multi-server clustering, and replicas.
- Unlimited CPUs
- CUDA GPU retrieval
- Entropy worker
- Parallel servers & replicas
- Named engineer support
Fully managed dedicated VMs. We host and operate LightningDB on Azure. Billed per GB of working set per month — cheaper than Neo4j Aura at $65/GB. Enterprise is custom quoted.
Pro
$55 / GB / mo
Managed Pro. Dedicated Azure VM with 16 cores and 1 CUDA GPU.
- 16 CPUs
- 1 CUDA GPU retrieval
- Dedicated Azure VM
- Automated daily backups
- Business-hours support
Pro+
$85 / GB / mo
Managed Pro+. Dedicated Azure VM with 32 cores, 2 CUDA GPUs, and entropy worker.
- 32 CPUs
- 2 CUDA GPUs retrieval
- Entropy worker
- Dedicated Azure VM
- Business-hours support
Enterprise
Quote
Unlimited cores, multi-server parallel replication, named engineer support.
- Unlimited CPUs
- CUDA GPU retrieval
- Parallel servers & replicas
- Entropy worker
- Named engineer support
What is included.
Complete capability breakdown across all licensing and hosting tiers.
| Capability | Free | Pro | Pro+ | Enterprise |
|---|---|---|---|---|
| Queries and GraphRAG | Yes | Yes | Yes | Yes |
| CPUs | 2 | 16 | 32 | Unlimited |
| CUDA GPUs | No | 1 GPU | 2 GPUs | Multi-GPU / Unlimited |
| Entropy worker | No | No | Yes | Yes (Clustered) |
| Parallel servers | No | No | No | Yes |
| Replicas | No | No | No | Yes |
| Self-hosted (Flat) | Free download | $20 / mo (annual) · $25 monthly | $40 / mo (annual) · $50 monthly | Custom Quote |
| IP Lock & Porting | None | Single IP · Instant web porting | Single IP · Instant web porting | Multi-server clustering |
| Managed Cloud (Per GB) | — | $55 / GB / mo | $85 / GB / mo | Custom Quote |
| Support | Community | Business hours | Business hours | Named engineer |
Cloud vs Neo4j Aura.
Neo4j Aura list prices next to LightningDB Cloud on identical working-set volumes. Cheaper than Neo4j Aura ($65/GB); Enterprise Cloud is custom quoted.
| Working set | Aura Professional | Aura Business Critical | Cloud Pro ($55/GB) | Cloud Pro+ ($85/GB) | Cloud Enterprise |
|---|---|---|---|---|---|
| 8 GB | $520 | $1,168 | $440 | $680 | Quoted |
| 16 GB | $1,040 | $2,336 | $880 | $1,360 | Quoted |
| 32 GB | $2,080 | $4,672 | $1,760 | $2,720 | Quoted |
| 64 GB | $4,160 | $9,344 | $3,520 | $5,440 | Quoted |
Monthly figures in USD. Aura figures represent list price for hosted working sets. Cloud Pro and Pro+ are fully managed LightningDB at $55 and $85 per GB.
Frequently asked questions.
Do I need a GPU to beat Neo4j?
No. CPU-only LightningDB packs the 2.5 million node prompt in 9.18 ms, against 12.12 ms for Neo4j. On exact cosine the CPU lead is 46× to 630×. The GPU accelerates the large pack to 6.48 ms and the cosine ladder to 211×–1,830×.
What does node-locked mean?
Paid self-hosted editions ship as an optimized binary and a cryptographic license bound to one machine fingerprint. You send your hardware fingerprint, the license starts on that host, and copies elsewhere fail cleanly.
Is Free the same engine?
Yes. Identical CSR layout, identical query compiler, identical /v1/rag endpoint. It is capped at 2 CPUs and omits GPU and entropy workers. The open-source community proof script runs on it.
When does Cloud open?
We are currently onboarding from the waitlist. Tell us your target region, tier, and working set. Until regional clusters launch, Pro, Pro+, and Enterprise run on hardware you control.
Start free. Scale when the graph grows.
Free Edition is all you need to run local benchmarks and build your pipeline. Upgrade to Pro or Pro+ when you need CUDA acceleration and higher core counts.
# Quick check with Free Edition
./build/graphdb serve --port 7475
curl http://localhost:7475/v1/status