Benchmark what you have.
Then size the AI.
Reuse. Benchmark. Specialize. Escalate.
Tell us what hardware you already own and how your team expects to use AI. The tool estimates a practical starting model tier - including Liquid LFM2-class or specialist alternatives - and compares a local-first routing plan with a centralized foundation-model reference.
Run a planning estimate.
Use non-sensitive specifications only. Do not enter confidential architecture details, hostnames, IP addresses, customer data, credentials, export-controlled information, or classified information.
Your right-sizing snapshot.
The result below is generated from the submitted assumptions and the Bates benchmark reference values.
-
-
Environmental figures are scenario-planning proxies, not measured facility savings. Local building cooling, idle power, embodied hardware, network energy, provider batching, grid mix, and full data-center facility overhead are outside this calculator unless already represented by the comparison reference.