Share this article

Latest news

With KB5043178 to Release Preview Channel, Microsoft advises Windows 11 users to plug in when the battery is low

Copilot in Outlook will generate personalized themes for you to customize the app

Microsoft will raise the price of its 365 Suite to include AI capabilities

Death Stranding Director’s Cut is now Xbox X|S at a huge discount

Outlook will let users create custom account icons so they can tell their accounts apart easier

OpenAI fine-tunes its GPT-4o and GPT-4o Mini, making the already affordable models even more competent

It scored 82% on the MMLU benchmark.

2 min. read

Updated onAugust 21, 2024

updated onAugust 21, 2024

Share this article

Read our disclosure page to find out how can you help Windows Report sustain the editorial teamRead more

OpenAI announced its newest AI model, GPT-4o Mini, a more affordable version of the latest GPT-4o model, to make access to advanced artificial intelligence more democratic. Although the costs of using this new model are lessened, it still performs well, strengthening its competition in the field.

However, it seems the company is taking things even further: OpenAI has rolled outfine-tuning support for its GPT-4o and GPT-4o mini models. This is a game-changer for those looking to tailor AI responses to specific needs, making the technology not only more accessible but also more versatile. With this new feature, developers can train these models using custom datasets, which promises higher performance at a lower cost for particular applications.

OpenAI isn’t holding back on making this feature accessible, either. They’re offering a generous amount of training tokens for free every day to every organization—1 million for GPT-4o fine-tuning and 2 million for GPT-4o mini. This offer, available through September 23, invites developers to dive in and explore the possibilities. The cost for additional training and inference is surprisingly affordable, making it an attractive option for businesses of all sizes.

Getting started is straightforward. Developers head to the fine-tuning dashboard on the OpenAI platform, select their base model, and begin the customization process. This user-friendly approach lowers the barrier to entry for those new to AI development.

The results from early adopters are promising. Distyl, for instance, snagged the top spot on the BIRD-SQL benchmark thanks to its fine-tuned GPT-4o, showcasing the model’s prowess in SQL generation and other complex tasks. Genie, another early tester, achieved state-of-the-art scores on benchmarks, highlighting the significant improvements that fine-tuning can bring to AI performance.

This development clearly signals that OpenAI is committed to making its AI models powerful, adaptable, and easy to use for developers. Whether you’re a seasoned AI expert or a curious newcomer, the potential for innovation and customization is now at your fingertips. So, what could you create with a fine-tuned GPT-4o or GPT-4o mini at your disposal?

More about the topics:AI,OpenAI

Claudiu Andone

Windows Toubleshooting Expert

Oldtimer in the tech and science press, Claudiu is focused on whatever comes new from Microsoft.

His abrupt interest in computers started when he saw the first Home Computer as a kid. However, his passion for Windows and everything related became obvious when he became a sys admin in a computer science high school.

With 14 years of experience in writing about everything there is to know about science and technology, Claudiu also likes rock music, chilling in the garden, and Star Wars. May the force be with you, always!

User forum

0 messages

Sort by:LatestOldestMost Votes

Comment*

Name*

Email*

Commenting as.Not you?

Save information for future comments

Comment

Δ

Claudiu Andone

Windows Toubleshooting Expert

Oldtimer in the tech and science press, with 14 years of experience in writing on everything there is to know about science, technology, and Microsoft