Choosing a model

Should your support AI run locally or in the cloud?

ZChat can drive its chatbot and reply suggestions from a local Ollama model on your own hardware, or from OpenAI or Anthropic with your own API key. The right answer depends on what your conversations contain and what you're willing to run. Here's the trade-off without the sales pitch.

Local model (Ollama) Hosted (OpenAI / Anthropic)
Where conversations go Nowhere - they stay on your server To the provider, under their terms
Cost shape Hardware you already pay for Per token, scales with volume
Answer quality Good, and improving fast Currently the strongest available
Setup effort Install a runtime, pull a model Paste an API key
Works offline / air-gapped Yes No - needs outbound internet
Who you depend on Yourself, including uptime A vendor's availability and pricing

The thing most comparisons get wrong

Model benchmarks measure open-ended reasoning. Support chat mostly isn't that. The answers your customers need are in your refund policy, your setup steps, your compatibility table - facts no model knows regardless of size.

That's why grounding the AI in your own documents usually improves answers more than upgrading the model does. A smaller local model with good documentation will routinely beat a frontier model guessing at your product.

Where the bigger model genuinely wins: messy, ambiguous questions; customers writing in a second language; and following long multi-step conversations without losing the thread. If that describes your traffic, weight quality more heavily.

Choose local when

  • Conversations carry personal, health or financial data
  • A policy or regulator says data stays in your control
  • You run on an isolated network
  • Volume is high enough that per-token pricing stings
  • You want a cost that doesn't move with usage

Choose hosted when

  • You want to find out today whether AI helps at all
  • Nobody wants to operate another service
  • You have no suitable hardware
  • Your questions are open-ended or multilingual
  • Volume is low enough that the bill stays small
It's a setting, not a commitment

You are not locked in either way

The provider is configuration. Start hosted to prove the AI earns its place on your traffic, then move local once you know it does - or start local because policy requires it and never look back. Either way ZChat itself adds no per-message charge; a hosted provider bills you directly on your own key.

Is a local model as good as a hosted one?
Generally not, at the same moment in time. The largest hosted models are still stronger at open-ended reasoning. For support chat the gap matters less than people expect, because the useful answers come from your own documentation rather than the model's world knowledge - but if you need the strongest possible reasoning, a hosted model is still the stronger pick.
What does a local model actually need to run?
A machine with enough memory for the model you choose, ideally with a GPU. Small models run acceptably on modern CPUs; larger ones want dedicated hardware. This is a real cost - it is your server rather than a per-message bill, but it is not free.
Can I change my mind later?
Yes. The provider is a setting, not an architecture. You can start on a hosted model to see whether AI helps your team at all, then move to a local model once you know it earns its place - or the reverse.
Does ZChat charge per AI message?
No. There is no metering or credit system in ZChat itself. With a hosted provider you pay that provider directly using your own key; with a local model there is no per-message cost at all.
Which should most teams start with?
If your conversations contain personal or regulated data, start local - that decision is made for you. Otherwise start with whichever you can set up fastest, prove the AI is actually useful on your traffic, and only then optimise for cost or privacy.
Owned customer support software

Deploy live chat on your own terms, not on someone else's pricing model.

ZChat gives you the installable server, web dashboard, website widget, and desktop agent tools in one self-hosted product you buy once and keep. Run it on infrastructure you trust and connect AI only if and how you want it.

Deployment

Install on Windows or Linux, behind IIS or Nginx, in a VM, or in Docker if that fits your stack.

Commercial model

One-time purchase, perpetual license, and no monthly per-agent bill attached to growth.

AI Flexibility

Use Ollama locally or connect OpenAI and Anthropic with your own provider accounts.