GPT-5 (low) is a specific configuration of OpenAI’s flagship GPT-5 model, optimized to balance computational speed with logical reasoning depth. Released on August 7, 2025, it represents a middle ground in the GPT-5 series' reasoning effort settings, designed for tasks that require more analytical consistency than standard completion models but lower latency than "high" reasoning modes. It is part of a multimodal family that supports both text and image inputs.
The model features a context window of 400,000 tokens, enabling the processing of extensive technical documentation and long-form conversational histories. GPT-5 (low) utilizes a hierarchical routing system and reasoning checkpoints—similar to chain-of-thought introspection—to validate intermediate steps before generating a final response. This architecture contributes to a significant reduction in hallucinations compared to the GPT-4 series, particularly in specialized domains such as mathematics, programming, and physics.
In benchmark evaluations, GPT-5 (low) has demonstrated high reliability in structured data generation and complex instruction following. While it generates more tokens per second than the higher-effort variants, it maintains the core capabilities of the GPT-5 architecture, including improved tool-use accuracy and enhanced agentic behavior for multi-step workflows.