Cloud AI Pricing & Plans

Questions

Why is it priced this way?

Plans use one combined monthly allowance. Consumption depends on the model, input/output tokens, caching, and the applicable policy, so the same allowance can cover different amounts of usage across models.

For text PAYG, published prices follow the confirmed developer Standard/list rate. Media follows the verified rate for the unit shown.

How does BotConnector make money?

Plan fees cover AI usage, payment processing, support, and BotConnector infrastructure. Procurement cost can be lower when qualified routes, more efficient infrastructure pricing, prompt caching, or provider credits and promotions are available. That procurement efficiency benefits BotConnector without changing the published PAYG customer price.

Can prompt caching make usage more efficient?

BotConnector uses official prompt or context caching mechanisms available on supported models and providers. BotConnector does not return old AI answers as a response cache.

When a provider reports a cache hit or cache read, that portion of the input may be metered more efficiently according to the applicable model rate and policy. A cache hit is not guaranteed on every request.

Why isn't there one fixed token limit?

Models have different costs, so the same monthly allowance can cover more requests on efficient models than on premium models.

What is the difference between Starter, plans, and PAYG?

Starter provides access to available free cloud models under dynamic fair use and shared capacity.

Paid plans provide one combined monthly allowance for the models included in the plan. Available Starter free models remain accessible.

PAYG uses a separate prepaid balance based on actual usage for models and media available through PAYG.

What happens when plan usage is exhausted?

Included models stop drawing from the monthly allowance until the next period. PAYG models can still be used with PAYG balance when available.

What about Local AI?

Local AI inference runs on the user's own hardware and does not use the Cloud AI allowance. Account, pairing, model download, or other online services may still require a BotConnector connection.

Can BotConnector replace the model I select?

BotConnector keeps the model selected by the user. If the same model is available through more than one qualified route, the provider route may change based on availability and service conditions without replacing the selected model.

Why can the same model produce different answers?

AI models are generative. For open-weight models, provider runtime, hardware, precision, quantization, generation parameters, and tool-calling implementation can also affect output and performance.

What happens if a provider is slow or unavailable?

If a qualified alternative route exists for the same model, BotConnector may use it. If no suitable route is available, the request may fail rather than silently switching to another model. See the service status.

Where are my prompts and files sent?

For Cloud AI, requests pass through BotConnector to the vendor or infrastructure provider processing the selected model. The processing provider may differ between models and may change according to the active routing configuration.

For Local AI, model inference runs on the user's device. Details about BotConnector data handling are available in the Privacy Policy.

Is using a model through BotConnector the same as subscribing to its official app?

No. BotConnector provides access through supported APIs and infrastructure. Features specific to official consumer apps, such as particular workspaces, memory, integrations, or exclusive UI features, are not automatically included through BotConnector.

Refunds

See the refund policy.