How this channel uses the term In the channel's captured material, temperature appears as one of the sampling controls available when working with a language model directly. It is mentioned alongside stop tokens and system prompts as a setting exposed in the OpenAI playground — the raw interface for constructing requests to a model, where Notion AI research engineer Linus Lee applies his more advanced prompt-engineering techniques. [^e9ff2088-b9c3-51c2-8d93-456b3cc1bccf@4362] Where it fits The playground comes up in the conversation as the place for single-turn, product-grade prompting — techniques like chain-of-density drafting and putting words in the model's mouth via a fabricated chat history. Temperature is listed there as one of the dials you can turn when you want finer control over generation than the chat interface gives you. [^e9ff2088-b9c3-51c2-8d93-456b3cc1bccf@4362] The moment it appears: Lee is walking through advanced tricks from the playground and notes that it gives control over temperature, stop tokens, and system prompts — a reminder that behind the chat window sits a configurable model, not just a conversation. [^e9ff2088-b9c3-51c2-8d93-456b3cc1bccf@4362] What the evidence does (and does not) cover The material names temperature as a playground control but does not elaborate on its settings, values, or effects — so this entry stays deliberately narrow and does not import outside definitions. - Temperature appears in this channel's material as a model-level setting exposed in the OpenAI playground. [^e9ff2088-b9c3-51c2-8d93-456b3cc1bccf@4362]- It is grouped with stop tokens and system prompts among the controls used for advanced, single-turn prompt engineering. [^e9ff2088-b9c3-51c2-8d93-456b3cc1bccf@4362]- The captured material mentions the control without detailing how it works.