OpenAI modernizes enterprise applications | OnlineMarketing.de

We've introduced more enterprise-level features for API customers, including enhanced security, administrative controls, new Helpers API capabilities, and tools to help better manage costs. https://t.co/Ej8xxpML5T

– OpenAI (@OpenAI) April 23, 2024

Helper API improvements

The Help API, a core offering of OpenAI for businesses, has been significantly expanded. It now allows up to 10,000 files to be processed per assistant, a significant increase from the previous limit of 20 files. This scalability is complemented by new features such as parallel queries, improved reordering, and query rewriting, which improve the efficiency and accuracy of data preparation.

New security features and administrative controls

One of the key security improvements is the introduction of PrivateLink, a feature that enables secure and private communication between the OpenAI API and the Microsoft Azure Cloud. This is especially useful for companies that rely on high security standards. Additionally, OpenAI has introduced multi-factor authentication to increase access security.

The new Projects feature provides improved management visibility and control by allowing companies to manage roles and API keys at the project level. Users can now assign API keys and roles to specific projects, create allowlists and various OpenAI models, and set usage limits and fees for access to avoid overcharging for usage.

New cost management features

In the area of ​​cost management, OpenAI has introduced two new functions. On the one hand, this includes the reduced use of the promised throughput, where GPT-4 or GPT-4 Turbo clients using a certain number of tokens per minute can request access to the so-called “provided throughput” depending on the scope of their commitment to receive discounts ranging from ten And 50 percent. On the other hand, a batch API has been introduced that provides asynchronous and cost-effective processing options for less urgent requests and can thus save up to 50 percent of costs (if results are delivered within 24 hours). This might be interesting, for example, for investigative journalists looking through large amounts of documents for research purposes.

The updates, with a focus on security, administrative oversight and cost management, show that OpenAI aims to reduce complexity for businesses and developers and deliver technology that is ready to use out of the box.

Meta:

The new Llama 3 AI model is said to be better than competitors