AbacusAI has launched a new inference API called 'LLM Fine-Tunes', designed to enhance the performance of open-source large language models (LLMs) while being more affordable than competitors like GPT-4o. The API aims to provide businesses with advanced capabilities in natural language processing. Additionally, AbacusAI introduced its ChatLLM Teams feature, which allows users to interact with and analyze business data using state-of-the-art LLMs. Meanwhile, Microsoft has unveiled a new series of open-source models in its Phi-3.5 lineup, including Phi-3.5 Mini, MoE, and Vision. These models reportedly outperform larger models from Meta, Mistral, and Google in various AI tasks. OpenAI has also made its GPT-4o mini model available on Azure AI, which is designed for faster reasoning across text and vision at reduced costs.