What is a Temperature (AI)?

Temperature is a sampling parameter that controls how random or deterministic a language model's next-token choices are, with lower values favoring the highest-probability token and higher values spreading probability mass across more options.

Under the hood, a language model produces a probability distribution over its entire vocabulary for the next token. Temperature rescales that distribution before sampling: dividing the logits by a value below 1 sharpens the distribution toward the most likely tokens, producing more repeatable and conservative output, while a value above 1 flattens it, making less-likely tokens more competitive and output more varied. A temperature of 0 effectively becomes greedy decoding, always picking the top token.

Choosing temperature is a tradeoff between reliability and creativity. Low temperature suits tasks with a single correct answer, such as code generation, structured data extraction, or factual Q&A, where consistency matters more than variety. Higher temperature suits brainstorming or creative writing, where some randomness produces more useful variation across runs.

In practice with Neotask

Neotask's agent runtime keeps temperature low for tool-calling and data-extraction steps, so the same input reliably produces the same structured output, while creative tasks like drafting marketing copy or blog ideas run at a higher temperature to give the user several distinct options to choose from.

Related terms

• Top-p Sampling
• Tokens (AI)
• Tool Calling
• Large Language Model
• Prompt Engineering

Plans

Related terms

• Zero-Shot Learning - Zero-shot learning is the ability of a machine learning model to perform a task or classify inputs it has never seen labeled examples of during…
• Training Data - Training data is the corpus of examples, whether text, code, images, or other content, that a machine learning model learns statistical patterns…
• Agentic AI - Agentic AI refers to AI systems that autonomously plan, decide, and execute multi-step actions toward a goal, using tools and external state, rather…
• Reinforcement Learning - Reinforcement learning (RL) is a machine learning approach in which an agent learns to take actions in an environment by receiving rewards or…
• GPU Computing - GPU computing is the use of graphics processing units — chips originally built for rendering images — to perform massively parallel numerical…
• Fraud Detection Automation - Fraud detection automation is the use of rules engines and machine-learning models to automatically flag or block suspicious transactions and…

Explore: Integrations · Skills · Solutions · Use cases · Examples · Comparisons · Templates · Blog · Docs