Model Overview
GPT-4.5 Preview is a general-purpose GPT-4.5 research preview model released by OpenAI on February 27, 2025, with a focus on natural conversation, broad knowledge, creativity, and understanding user intent. It is not a reasoning model primarily characterized by extended reasoning before answering. (openai.com)
Capabilities and Uses
- Supports text input and output, image input, function calling, structured outputs, and streaming; it does not support audio or video. (developers.openai.com)
- Can be used for writing assistance, communication, learning, brainstorming, programming, complex task automation, and planning multi-step workflows. (openai.com)
- The context window is 128,000 tokens, and the maximum output is 16,384 tokens; the knowledge cutoff date is October 1, 2023. (developers.openai.com)
Limitations and Status
The model may still produce errors or hallucinations, and its output requires verification; it is large, computationally expensive, and does not support fine-tuning. (openai.com)
OpenAI announced the deprecation of this API model on April 14, 2025, and shut it down on July 14, 2025; the official model page also marks it as shut down, so its historical prices do not mean it is currently available for use. OpenAI officially recommends using GPT-4.1 or the o3 series for most purposes instead. (developers.openai.com)
Historical API Pricing
The standard text token prices published by OpenAI are in US dollars (USD): $75 per million input tokens, $37.50 per million cached input tokens, and $150 per million output tokens. Cached input pricing applies only to inputs eligible for caching; these are historical prices for a shut-down model, not currently available rates. Date the official model page was checked: November 19, 2026. (developers.openai.com)