OpenAI is once again making waves in the artificial intelligence landscape with their groundbreaking updates. In their latest blog post, “New embedding models and API updates,” OpenAI introduces exciting advancements that will reshape the world of natural language processing and machine learning. Among the OpenAI highlights are the upcoming release of a free GPT-4 Turbo model with vision capabilities and a significant reduction in pricing for the popular GPT-3.5 Turbo. Let’s delve into the details of these game-changing updates.
Table of Contents
Exciting Updates to GPT Models
OpenAI has introduced groundbreaking updates to their GPT models. Let’s explore what’s new in the world of GPT-4 Turbo and GPT-3.5 Turbo.
1. Lower Pricing for GPT-3.5 Turbo
OpenAI is committed to making their models accessible to a wider audience, and they are delivering on this promise by reducing the prices for GPT-3.5 Turbo. The upcoming GPT-3.5 Turbo model, gpt-3.5-turbo-0125, will see input prices slashed by 50% to $0.0005 per 1k tokens and output prices reduced by 25% to $0.0015 per 1k tokens. These price adjustments make GPT-3.5 Turbo even more affordable, empowering developers to scale their projects without breaking the bank.
2. Enhanced GPT-4 Turbo Preview Model
OpenAI’s GPT-4 Turbo has been gaining traction since its release, with over 70% of requests from GPT-4 API customers transitioning to GPT-4 Turbo. In response to user feedback, OpenAI is launching an updated GPT-4 Turbo preview model, gpt-4-0125-preview. This new version addresses previous issues of “laziness” and includes a fix for a bug affecting non-English UTF-8 generations. OpenAI is dedicated to refining their models to ensure optimal performance and user satisfaction.
| For full details about this update, please refer to our article: “OpenAI’s New Models To Counter Laziness of GPT-4 with GPT-4-0125 preview.”
A new model name alias, gpt-4-turbo-preview, is also coming, which will always point to the latest GPT-4 Turbo preview model. Users who want to stay up-to-date with the newest developments can rely on the gpt-4-turbo-preview alias to access the most recent version effortlessly.
3. Free GPT-4 Turbo With Vision
OpenAI shared exciting plans for the full launch of GPT-4 Turbo with Vision soon. They stated that over 70% of customers have now transitioned to the GPT-4 Turbo model since its release. Based on this strong customer response, OpenAI plans to launch GPT-4 Turbo with Vision in general availability in the coming months for absolutely free. By including the ability to understand images and videos in addition to text, GPT-4 Turbo with vision has the potential to be a true general intelligence. It holds tremendous potential to revolutionize the field of natural language processing and machine learning.
Introducing New Embedding Models
OpenAI is set to launch a new generation of embedding models. These new models include Text-embedding-3-small and Text-embedding-3-large that cater to different needs:
1. Text-embedding-3-small
The text-embedding-3-small model is a highly efficient embedding model that outperforms its predecessor, the text-embedding-ada-002 model. It offers stronger performance, with increased average scores on commonly used benchmarks for multi-language retrieval and English tasks. Additionally, it comes at a reduced price of $0.00002 per 1k tokens, making it a cost-effective choice for developers.
2. Text-embedding-3-large
The text-embedding-3-large model is the next generation of larger embedding models. With embeddings of up to 3072 dimensions, it provides even stronger performance compared to the previous text-embedding-ada-002 model. Despite its enhanced capabilities, OpenAI has managed to keep the pricing competitive at $0.00013 per 1k tokens.
New Improved Moderation Model
OpenAI places a strong emphasis on safety and has released text-moderation-007, their most robust moderation model to date. This model is designed to identify potentially harmful text and contribute to the creation of safer AI systems. Developers can utilize the free Moderation API to leverage this powerful tool and enhance the safety of their applications.
Native Support for Shortening Embeddings
OpenAI understands that using larger embeddings can be computationally expensive. To address this, developers can now shorten embeddings without sacrificing their concept-representing properties. This technique allows for more flexible usage and cost optimization, enabling developers to strike a balance between performance and resource consumption.
Exciting Updates for Developers
In addition to the groundbreaking embedding models and free GPT-4 Turbo access announcements, OpenAI is also introducing two platform improvements to provide developers with more control and visibility over their API usage and keys: API Key Permissions and Usage Dashboard and Export Function. Developers can now assign permissions to their API keys. OpenAI has also enhanced the usage dashboard and export function to provide detailed insights into API usage. Developers can now track key metrics, such as usage volume and types of API calls, to better understand and optimize their usage.
Positioning AI for an Exciting Future
In summation, OpenAI’s announcements signify the next level for general-purpose AI and language models. The tidal wave of possibilities these updates unleash is sure to drive the AI future in thrilling new directions. Free GPT-4 Turbo with vision and GPT-3.5 Turbo with low pricing are surely groundbreaking updates.
