--- title: "What a Model Costs" url: "https://docs.noreja.com/en/article/what-a-model-costs" language: "en" last_updated: "2026-10-23T17:09:09.739Z" --- # What a Model Costs Where you choose the model for Minerva, what the dollar signs and prices next to a model mean, when Noreja warns about premium models, and where you see what a chat has cost. Minerva can answer with a local model or with an external model from a cloud provider. External models are paid per token, the small pieces of text a model reads and writes, and their prices differ a lot. This page explains where you choose the model, how Noreja shows what a model costs, and where you see what a chat has cost. Which data goes to which model is explained in [Minerva: Models, Data Flow and Data Protection](https://docs.noreja.com/en/article/minerva-models-data-flow-and-data-protection). > **Permissions:** feature **Chat**: **Read** opens the chat page, where you can ask questions and choose the model; **Create** adds the **Minerva-AI** switch next to dashboards and the Analyzer. See [Create and Edit Roles](https://docs.noreja.com/en/article/create-and-edit-roles). ### Where You Choose the Model The model button below the input field shows the model that writes the next answer. Click it to choose another one. The menu groups the models: - **Local models**: run on servers operated by Noreja or, in an on-premise installation, by your company. They aren't paid per token and have no price range. - **External Models**: models of cloud providers such as Anthropic, OpenAI and Mistral, paid per token. You can switch the model during a chat. When you switch from a local to an external model, Noreja warns you first, because the next questions are then sent to the cloud provider. It doesn't warn when a chat starts on an external model; check the model before you ask something confidential. How to ask and work with the answers is explained in [Chat with Minerva](https://docs.noreja.com/en/article/chat-with-minerva). ### The Price Range of a Model Next to each external model, in the model menu and on the model button, Noreja shows its price range as dollar signs: - **One dollar sign**: one of the cheapest models. - **Two dollar signs**: a medium-priced model. - **Three dollar signs**: an expensive model. - **Three dollar signs and a plus**: one of the most expensive models. The signs that don't apply are shown in grey, so every model has room for three. ![The model menu with the local model Gemma 4 and the external models of Anthropic, OpenAI and Mistral, each with its price range from one to three dollar signs, two of them with a plus](https://cdn-assets.productfruits.com/84b12813-2d0c-40ae-a448-46d69f1c9cf8) ### The Prices per Token Hover over the dollar signs of a model in the menu, or over the model button for the selected model, to see its prices per one million tokens, in US dollars: - **Input**: the text sent to the model, that is your question, the earlier conversation and the data Minerva reads. - **Output**: the text the model writes. - **Cached input**: input the provider still has from an earlier request in the chat; it is charged at this lower price. ![The model button with GPT 6.1 Sol and its price range, and the tooltip with its prices per 1M tokens for input, output and cached input](https://cdn-assets.productfruits.com/e91ad733-1712-46ae-9a95-4ea852bbe637) Long chats cost more than short ones: with every question, the model reads the conversation so far again. For a new topic, start a new chat. ### Premium Models When you choose one of the most expensive models, Noreja asks first (**Premium model selected**) and names its prices for input and output. Click **Confirm** to use it, or **Cancel** to keep the current model. With **Do not show this warning again**, Noreja stops asking until you close the browser tab. ### What a Chat Has Cost - **Below each answer**: how many tokens the models used for it, with an estimated cost. - **Token usage** (next to the model button): opens **Conversation Tokens**, the tokens the whole chat used by model, agent and tool, with an estimate of the cost. The costs are estimates, calculated from the tokens and the model's prices. ### Models for Agents Whoever sets up agents in the AI Hub chooses which models an agent may use. The same price ranges appear there. See [AI Hub: Agents, Skills, Prompts and Workspaces](https://docs.noreja.com/en/article/ai-hub-agents-skills-prompts-and-workspaces). ### Related - [Chat with Minerva](https://docs.noreja.com/en/article/chat-with-minerva) - [Minerva: Models, Data Flow and Data Protection](https://docs.noreja.com/en/article/minerva-models-data-flow-and-data-protection) - [Get Better Answers from Minerva](https://docs.noreja.com/en/article/get-better-answers-from-minerva) - [Minerva FAQ](https://docs.noreja.com/en/article/minerva-faq)