GPT 4.5
GPT-4.5 was a large non-reasoning model from OpenAI, released as a research preview on February 27, 2025 and removed from the API on July 14, 2025. OpenAI scaled it through pre-training rather than inference-time reasoning, and recommended GPT-4.1 as the replacement when it retired the preview. OpenAI did not publish its parameter count or architecture, describing it only as its largest model to date and pointing to broader world knowledge, better instruction following and a lower hallucination rate than GPT-4o on its own evaluations. Alignment relied on reinforcement learning from human feedback and on the behavior rules set out in OpenAI's published Model Spec. In OpenAI's own framing the gains were in writing, general knowledge and conversational nuance rather than in the mathematics and coding benchmarks where its reasoning models scored higher.
While it was available, it was used for writing assistance, research support, content automation and customer-facing systems; its price per token was far higher than GPT-4o's, which was one reason OpenAI withdrew it from the API. It is no longer available for new deployments; OpenAI pointed developers to GPT-4.1 in its place.
Related terms
Related services: LLM development, Agentic AI consulting.
Ready to put agentic AI to work?
Book a free 45-minute consultation. We'll map one real process worth automating with production-grade AI.