← Back to changelog

February 27, 2025

OpenAI GPT-4.5 support

Marc Klingen

Langfuse now offers full support for OpenAI's GPT-4.5, featuring day-one integration with our LLM playground, LLM-as-a-judge evaluation system, and comprehensive cost tracking capabilities.

OpenAI has released GPT-4.5, their largest and best model for chat yet. This research preview represents a significant advancement in scaling up pre-training and post-training capabilities.

Core features include (from OpenAI release notes):

• Scaled Unsupervised Learning: GPT-4.5 focuses on scaling unsupervised learning, improving its ability to recognize patterns, draw connections, and generate creative insights without reasoning.
• Enhanced “EQ” and Reduced Hallucinations: Early testing shows GPT-4.5 feels more natural to interact with, featuring a broader knowledge base, improved ability to follow user intent, and greater emotional intelligence. It’s also expected to hallucinate less.
• Deeper World Knowledge: The model demonstrates significantly improved factual accuracy and knowledge depth compared to previous models, with benchmark tests showing substantial improvements over GPT-4o.
• Natural Conversation: GPT-4.5 integrates ideas naturally in warm and intuitive conversations that are more attuned to human collaboration, with better understanding of what humans mean and interpretation of subtle cues.

GPT-4.5 is a very large and compute-intensive model, making it more expensive than GPT-4o. Thus, cost tracking is essential when using it in production.

Try it out

• Langfuse LLM playground
• Langfuse LLM-as-a-judge
• Langfuse model usage and cost tracking