ReasoningRequestParameter
Enums · AiDotNet.Enums
Names one per-request chat setting so a reasoning-model profile can declare it unsupported.
For Beginners
When you ask an AI for a reply you can attach settings such as “be more creative” (temperature) or “stop after 500 words” (maximum output tokens). Newer reasoning models do not accept all of them. This list simply gives each setting a name so the library can report, in plain terms, which one it had to leave out of your request and why.
How It Works
Reasoning models reject some of the sampling controls that ordinary chat models accept. Sending
temperature or top_p to an o-series or GPT-5 model is answered with an HTTP 400, so the request
has to be adjusted before it leaves the process rather than after the provider rejects it. This enumeration is
the vocabulary a ReasoningModelProfile uses to say which settings must be removed, and the vocabulary a
diagnostic uses to say which setting it removed.
The members name the provider-neutral ChatOptions properties rather than the wire fields, because the
same profile has to drive several connectors whose wire names differ. The mapping to a wire field is the
connector’s job.
Fields
| Field | Summary |
|---|---|
ReasoningRequestParameter MaxOutputTokens |
The answer length cap (ChatOptions.MaxOutputTokens, wire field max_tokens). |
ReasoningRequestParameter ReasoningEffort |
The deliberation level a reasoning model accepts (wire field reasoning_effort). |
ReasoningRequestParameter ResponseFormat |
The structured-output request (ChatOptions.ResponseFormat, wire field response_format). |
ReasoningRequestParameter Seed |
The deterministic sampling seed (ChatOptions.Seed, wire field seed). |
ReasoningRequestParameter StopSequences |
The stop sequences (ChatOptions.StopSequences, wire field stop). |
ReasoningRequestParameter SystemMessageRole |
The role name system messages are sent under (wire field role on the first message). |
ReasoningRequestParameter Temperature |
The sampling temperature (ChatOptions.Temperature, wire field temperature). |
ReasoningRequestParameter Tools |
The tool declarations (ChatOptions.Tools, wire fields tools and tool_choice). |
ReasoningRequestParameter TopK |
Top-K sampling (ChatOptions.TopK); most OpenAI-shaped endpoints never send it. |
ReasoningRequestParameter TopP |
Nucleus sampling mass (ChatOptions.TopP, wire field top_p). |
int value__ |