Pricing
Simple, transparent pricing.
Start free. Pay only for what you compute.
Free
For exploration and small experiments.
- ✓ 10 jobs per day
- ✓ 1,024 neurons max
- ✓ 1,000 timesteps max
- ✓ 30s max runtime
- ✓ Community support
Researcher
For academic research and coursework.
- ✓ 100 jobs per day
- ✓ 32,768 neurons max
- ✓ 100,000 timesteps max
- ✓ 5 min max runtime
- ✓ Email support
- ✓ Compute: £18/hr
Startup
For product development and benchmarking.
- ✓ 500 jobs per day
- ✓ 131,072 neurons max
- ✓ 1M timesteps max
- ✓ 30 min max runtime
- ✓ Priority support
- ✓ Compute: £14.40/hr
Enterprise
For production workloads and dedicated capacity.
- ✓ Unlimited jobs
- ✓ 131,072 neurons max
- ✓ 10M timesteps max
- ✓ 1 hour max runtime
- ✓ Dedicated support
- ✓ Compute: £10.80/hr
How compute pricing works
You're billed for actual simulation time, not wall-clock time. A 1,000-neuron network running 10,000 timesteps might take 0.5 seconds of compute.
Example: Small experiment
100 neurons, 1,000 timesteps
Compute: ~0.05 seconds
Cost: £0.00025
Free tier: £0.00
Example: Full research run
32K neurons, 100K timesteps, 50 trials
Compute: ~300 seconds total
Cost: £1.50 (Researcher tier)
Less than a coffee.
Questions
Is the free tier actually free?
Yes. 10 jobs per day, up to 1,024 neurons, forever. No credit card required.
What neuromorphic model do you run?
Hardware-accurate Loihi 2 parity. Leak-integrate-fire neurons with 4 dendritic compartments, configurable thresholds, refractory periods, synaptic delays, and multiple learning rules including STDP and 3-factor reward modulation.
Can I export my network?
You can re-submit the same JSON definition at any time. The network definition is yours. The simulation engine runs server-side.
What's the latency?
Small simulations complete in under a second. Larger ones scale linearly with neuron count and timesteps. Results are available via polling.