PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteiTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Not as a fixed charge just for opening a session. In OpenAI’s Responses API MCP integration, you pay for tokens used to import tool definitions and make tool calls; OpenAI says there is no additional fee per tool call. Definitions can occupy request context while they remain available, however, and the documentation does not promise that retained definitions are token-free. The distinction is between fetching a tool list again and the token use associated with tool definitions in context.
What OpenAI says you pay for
OpenAI’s MCP servers guide states: “When you’re using the MCP tool, you only pay for tokens used when importing tool definitions or making tool calls. No additional fees apply per tool call.” This describes token usage in the Responses API MCP integration, not a separate fixed fee for every session or an additional fee each time a tool is called.
When you specify an MCP server in the Responses API tools parameter, the API attempts to retrieve that server’s tool list. If retrieval succeeds, the response includes an mcp_list_tools item containing the imported tools. Importing definitions can therefore add material to the request context, even if a particular request does not invoke every listed function.
Quick wins for a faster PC:
Scan for outdated or missing drivers - takes under a minuteDriver Scan →Clear out junk files and repair common Windows errorsFree Scan →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Does the API reload schemas on every turn?
Not while the mcp_list_tools item remains in the conversation context. OpenAI says the API does not fetch the tool list from the MCP server again at each turn under that condition. That is a statement about avoiding another remote fetch; it is not a statement that definitions retained in context consume no tokens. The documentation does not specify a zero-token treatment for retained definitions.
#1 Best Overall
So “billed on every session, used or not” is too broad: the cited guide explains token charges for importing definitions and making calls, but does not describe a fixed fee merely for opening a session with an unused MCP server. What definitions cost in token usage depends on what is imported and included in context; no universal token count or dollar estimate is established by the cited guidance.
How to limit unnecessary tool-definition overhead
Choose a loading approach based on how many tools your requests need and how often they need them. OpenAI’s tool search guide distinguishes eager loading from deferred loading:
| Approach | When it may fit | Trade-off |
|---|---|---|
| Eager loading | A small set of functions, or functions needed for most tasks. | Definitions are readily available, but unused definitions can occupy context. |
| Deferred loading | A large catalog where a task is likely to need only a subset. | Can reduce the amount of definition material loaded, but requires discovery and selection of the needed function. |
Deferred loading makes functions available only when needed and can reduce overall token usage. For a large catalog, tool search can help locate relevant functions rather than loading every definition up front. A separate OpenAI Cookbook guide also describes using allowed_tools to limit the tools exposed to a request. It presents this as a way to manage definition overhead, response time, and the model’s decision space—not as a guaranteed saving of a particular size.
How to decide whether a change helps
Compare the approaches on the workload you actually run. OpenAI recommends evaluating task completion, input-token usage, and latency. For a deferred or filtered setup, also check whether discovery reliably finds the function required by each task. Limiting the available definitions may reduce context overhead, but a selection strategy that misses a needed tool can undermine task completion. Do not infer a dollar saving from fewer definitions alone; measure token usage for your workload and apply the relevant model rates.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Where this answer applies
These billing and loading statements concern OpenAI’s Responses API MCP implementation. They should not be generalized to every MCP client, API provider, or separately billed host: those products may have different billing terms. The OpenAI guides cited here do not state a geographic limitation for this billing explanation.
Quick Recap
Best Value
Rank #4
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
