Buyer education · four kinds of product
A unified AI platform puts models from several providers behind one login or one API. That single phrase covers four products with different jobs, prices and risks. Pick the kind first, then the product. The test: if switching from one provider's model to another means a second login, a second bill and uploading your files again, you are not using one.
Disclosure. ZeroTwo publishes this guide and sells a product in the first kind. It does not score or rank products and leaves competitors' prices, model counts and limits out, because they change and we cannot keep them current. What it says about other products comes from their own pages, listed below with the date we read them. For our own figures, see the pricing page.
Decide the kind first
Stop at the first question you answer yes to. The order matters: the earlier questions are the more specific needs.
Look at a router or API gateway. ZeroTwo compared with OpenRouter.
Look at a bring-your-own-key client. ZeroTwo compared with TypingMind.
Look at a customer-facing agent builder.
A chat workspace is the fit. Choose by plan limits, the features you cannot give up and where your data goes. What to check before consolidating subscriptions.
The four kinds
Examples, in their own words
We read each vendor's own page on 5 October 2026 and describe only what it says. Plans, prices and limits are not repeated here: check each vendor's page for the current ones. ZeroTwo is the fifth example of the first kind: a multi-model workspace with 60+ models.
| Product | Kind | What its own page says | Source (read 5 Oct 2026) |
|---|---|---|---|
| OpenRouter | Router | Describes itself as access to hundreds of AI models through a single API endpoint, and documents using the OpenAI SDK pointed at its base URL. | openrouter.ai/docs/quickstart |
| Poe | Chat workspace | Presents a consumer chat app with private and group chats and a long list of models and bots to choose from. | poe.com/about |
| TypingMind | Bring-your-own-key client | Describes a chat front end where you use your own API keys and pay for what you use, sold with a licence. | typingmind.com |
| Chatbase | Agent builder | Describes AI agents for customer support and sales across chat, email and voice, built from your own data sources. | chatbase.co |
Pay-per-token or subscription?
A fixed "conversations per week" rule of thumb ignores how long your conversations are and which model you call. Enter your own numbers. The token prices are placeholders: replace them with the current prices on the provider's page.
Cost = conversations × turns × (input tokens × input price + output tokens × output price) ÷ 1,000,000. The break-even is the plan price divided by the cost per conversation.
Read the result with care. A plan is not an unlimited token supply: ZeroTwo's plans carry credit and request allowances, listed on the pricing page, and a pay-per-token estimate leaves out the interface, files, search and your own time. Heavy, long-context work moves the break-even a long way, in both directions.
Privacy
Blanket privacy badges hide the differences that matter. Provider, plan, retention and optional features each change the answer, so ask these of every product, including ours. ZeroTwo's own terms are in its privacy policy and terms.
Often two parties: the platform and the model provider behind the request. Read both privacy policies.
Retention and training defaults can differ between free, individual and business plans. Find the terms for the plan you would be on.
Look for retention and deletion language in the policy, and for how to delete your account and chats.
Web search, connectors, memory and file tools can involve other services. Check each one you plan to use.
For teams: ask who can see conversations, export data or switch features off, and where that is documented.
Our position
Questions
A unified AI platform gives you models from more than one provider through a single account or a single API. The phrase covers four different kinds of product: chat workspaces, API routers, bring-your-own-key clients and customer-facing agent builders. They solve different problems, so decide which kind you need before you compare products.
An aggregator in the consumer sense is an app: it builds the chat, history and file features and lets you pick a model behind them. A router is infrastructure: it gives developers one endpoint and key that reach many models, and leaves the interface, memory and cost control to you. If you are asking this, ask a second question: are you building software, or using AI yourself?
Yes, that is what a chat workspace is for. ZeroTwo lists 60+ models from providers including OpenAI, Anthropic, Google, xAI and DeepSeek, and the current list is on the models page. Which models you can use depends on your plan. Other products in the same category exist; read their own plan pages for what they include.
It depends on your workload, not on a fixed number of conversations. Multiply conversations, turns, and tokens per turn by the provider's per-token prices, then compare the result with the plan price. The calculator on this page does that arithmetic with your own inputs. A subscription also carries limits and allowances, so compare what you would actually use, not only the headline price.
Only if you are building software or an automation that calls models. If you are the one chatting, drafting or analysing, an app is the simpler route. If you are building, a router or the provider's own API fits better, and our comparison with OpenRouter sets out where a chat workspace stops being the right tool.
This guide does not claim one. If you need programmatic access, check the ZeroTwo docs or ask support before you plan around it. For API-first work, a router or the model provider's own API is built for that job.
There is no category-wide answer. Privacy depends on the platform, the model provider behind each request, and your plan, which is why this page gives you five questions to ask instead of a badge. ZeroTwo's practices are in its privacy policy; read it together with the data terms of each provider whose models you use.
It depends on what the team does. A team that wants a shared chat workspace with several providers' models should compare workspace plans; ZeroTwo's Business plan is $39.99 per seat per month ($35.99 billed annually) with pooled credits. A team that is building a product needs an API, and a team that wants a customer-facing bot needs an agent builder.
Edition note: rewritten in October 2026. The earlier ten-product matrix was removed because its prices and feature flags could not be verified. Spotted an error? Email hello@zerotwo.ai.
Keep going
The Free plan needs no card. Read the current plan terms before you decide, or see ZeroTwo for teams.