> ## Documentation Index > Fetch the complete documentation index at: https://docs.trynito.ai/llms.txt > Use this file to discover all available pages before exploring further. # Pricing > How per-model cost works when you run Nito commands, and where to read the live rates. Nito prices your usage per model, and the rates are configured live. Each model's catalog entry shows its current rate, so the authoritative figures come from browsing the catalog rather than from any table printed here. ## How Pricing Works Usage is priced mostly per token, with separate rates for input and output, plus additive rates for cached input tokens, reasoning output tokens, and server-side tools such as web search and fetch. Before each command runs, Nito places a hold sized to the expected cost, then settles it against what you actually used. Failed requests are not billed. See [Checking Usage](/plans/checking-usage) for your balance. ## What Gets Billed A single call can accrue charges across several rate categories, each priced separately: | Category | Priced by | Notes | | ------------------- | --------- | --------------------------------------------------------------------------------- | | Input tokens | Per token | Separate rate from output. | | Output tokens | Per token | Separate rate from input. | | Cached input tokens | Per token | Cache read and cache write are metered separately from ordinary input. | | Reasoning output | Per token | Reasoning tokens are billed on top of ordinary output. | | Server-side tools | Per use | Web search, web fetch, and similar hosted tools bill in addition to model tokens. | Input and output are always separate rates; cache, reasoning, and server-tool charges are additive on top. If a model shows no rate for an item, that item is simply not configured for it, not that it is free. ## Where to Read Per-Model Cost Rates are configured live and vary by model, so there is no canonical static price table on this page. When you browse the catalog with the `models` command, each model's entry shows its per-token cost, so your estimate tracks the live configuration. Because size, speed, and cost vary between models, do not assume two models cost the same: check the catalog entry for the one you plan to use. ## Server-Side Tool Pricing Server-side tools such as web search and web fetch are billed in addition to model tokens, at their own per-use rates. Those rates are shown alongside the model on the catalog and are read live. See [Server-Side Tools](/features/server-side-tools) for what these tools do. ## Plan Pricing Is Separate Per-model cost is distinct from your plan. What your plan includes and what upgrading unlocks are covered in [What Paid Unlocks](/plans/what-paid-unlocks). For free plan limits, see [Free Usage](/plans/free-usage). ## Common Issues * **Assuming prices are fixed.** Rates are configured live and can change. Read them from the catalog when you browse. * **Expecting a rate on every model.** A rate appears only when configured for that model; absence means it is not yet set, not that the model is free. * **Forgetting tool and reasoning costs.** Web tools and reasoning tokens are billed on top of base input and output. * **Confusing per-model cost with your plan.** They are separate; see [What Paid Unlocks](/plans/what-paid-unlocks). ## Related Resources Browse every model in the catalog. Check your consumption and remaining allowance. What web search and fetch do, and how they bill.