Benchmarks

Measure. Reproduce. Publish.

We publish only numbers we measure ourselves, and we say how to reproduce them. Where a figure has not been measured under a documented method yet, we say that instead of estimating.

Local inference vs. the usual alternatives

This compares deployment models, not specific vendors — and it is a cost and compute comparison, not a capability benchmark. Large hosted models remain more capable on many tasks than a laptop-scale local model.

MetricNC AI (local CPU)Hosted AI subscriptionHosted AI, pay-per-tokenSelf-hosted open-weight model
Recurring costNoneTypically a monthly seat feeScales with usageNo licence fee; GPU hardware required
Hardware requiredAny CPU — no GPUNone (vendor-managed)None (vendor-managed)Datacenter-class GPU
Where data goesStays on your machineSent to the providerSent to the providerStays on your infrastructure
Works offlineYesNoNoYes, after download
Credentials neededNoneAccount + payment methodAPI key + payment methodNone
Runtime footprint570KB binaryN/A — hostedN/A — hostedMulti-GB model files
Why no vendor names or head-to-head numbers? Third-party pricing and performance change constantly, and quoting them turns a technical page into a marketing claim we would have to keep re-verifying. The comparison above describes deployment models, which is the decision teams actually make. Our own figures are the ones we publish and stand behind.
$0
NC AI — recurring cost
  • ✓ No per-seat or per-token billing
  • ✓ Runs on any CPU
  • ✓ Data stays local
  • ✓ No internet required
Per seat
Hosted subscription
  • ✗ Usually rate limited
  • ✗ Data leaves your network
  • ✗ Requires connectivity
Per token
Hosted API
  • ✗ Cost scales with usage
  • ✗ Provider dependency
  • ✗ Requires connectivity

Figures for NC AI are measured on a consumer laptop CPU (ARM64) using the runtime's own benchmark command. Reproduce them yourself: nc ai benchmark. Characteristics listed for other deployment models are general properties of those models, not measurements of any particular product.

By the Numbers

80+
NC Functions
570KB
Binary Size
0
Runtime Dependencies
113
Test Files Passing

We haven't yet published a formal Environment / Hardware / Workload / Reproduction-instructions writeup for every number on this site. As that documentation is finished, it will be linked from here rather than replacing the summary numbers above.