OpenAI recently shared several updates designed to enhance the experience for developers utilizing its AI services. These improvements encompass the introduction of new embedding models, a reduced cost for GPT-3.5 Turbo, an updated preview of GPT-4 Turbo, and enhanced content moderation capabilities.
The AI lab, headquartered in San Francisco, highlighted the superior performance of its recently introduced text-embedding-3-small and text-embedding-3-large models compared to their predecessors. Notably, text-embedding-3-large achieved an average score of 54.9 percent on the MIRACL benchmark and 64.6 percent on the MTEB benchmark. These figures represent a significant improvement over the older text-embedding-ada-002 model, which scored 31.4 percent and 61 percent on the respective benchmarks.
Furthermore, OpenAI disclosed a noteworthy reduction in the price per 1,000 tokens for text-embedding-3-small, dropping from $0.0001 to $0.00002 in comparison to text-embedding-ada-002. Developers can also curtail embeddings to minimize costs without compromising accuracy significantly, according to the company.
In the upcoming week, OpenAI plans to unveil an updated version of the GPT-3.5 Turbo model, accompanied by a 50 percent reduction in pricing for input tokens and a 25 percent reduction for output tokens. This marks the third price adjustment for GPT-3.5 Turbo in the past year as OpenAI seeks to encourage wider adoption of its services.
Additionally, OpenAI provided insights into the latest version of its GPT-4 Turbo preview, denoted as gpt-4-0125-preview. The company reported that over 70 percent of requests have transitioned to this model since its debut, with improvements noted in areas such as more comprehensive code generation and task completion.
Addressing the importance of building safe AI applications, OpenAI introduced its most advanced content moderation model, text-moderation-007. The company claimed that this model exhibits improved accuracy in identifying potentially harmful text compared to its predecessors.
Finally, developers now have increased control over API keys and greater visibility into usage metrics. OpenAI emphasizes that developers can assign specific permissions to keys and monitor consumption at a per-key level, facilitating more precise tracking of individual products or projects.
OpenAI has outlined plans for additional platform enhancements in the coming months, specifically tailored to accommodate larger development teams.
