OpenAI's first DevDay has already come and gone, and during this event, the company shared announcements and developments within ChatGPT and OpenAI with the public. If you missed the event, don't worry, we'll bring you up to speed with this summary of the most important highlights.
Huge milestone for ChatGPT
One of the most impressive updates is the enormous success of ChatGPT. Within a year of its launch, the tool has gained enormous popularity, with over 100 million users interacting with it weekly. Even more remarkable is that it reached the milestone of 100 million monthly users in just a few months, making it one of the fastest-growing consumer products.
In addition, more than 2 million developers actively use OpenAI's API to build innovative solutions.
Introducing GPT-4 Turbo
OpenAI unveiled the long-awaited GPT-4 Turbo during the event. This improved version of the GPT-4 model comes in two variants—one for text analysis only and the other for understanding both text and images. The text-only model costs $0.01 per 1,000 input tokens and $0.03 per 1,000 output tokens. For image processing, GPT-4 Turbo costs $0.00765 for a 1080×1080 pixel image.
With a context window of 128,000 tokens, approximately four times more than GPT-4, and a knowledge cutoff point extended to April 2023, GPT-4 Turbo promises improved capabilities and broader context.
Enabling users to build their own GPTs
OpenAI announced that users can now create their own versions of GPT for personal or productivity use, without any programming knowledge. This feature allows users to build bots through simple prompts and connect them to databases or knowledge bases, seamlessly importing external information.
In addition, OpenAI is launching a GPT Store later this month, where users can publish their creations. Initially featuring contributions from "verified builders," this store opens up new possibilities, with Altman mentioning potential payments for creators with popular GPTs.
New API for building assistants
OpenAI introduced the Assistants API, which allows developers to create their own "agent-like experiences." With this API, developers can create agents that can retrieve external knowledge or call programming functions for specific actions. The possibilities range from code assistants to AI-powered vacation planners, demonstrating the versatility of this new tool.
DALL-E 3 API and Text-to-Speech APIs
OpenAI expanded its offering with the DALL-E 3 API, which provides access to its text-to-image model with built-in moderation tools. Pricing is $0.04 per image generated, with output specifications ranging from 1024×1024 to 1792×1024 in various formats.
In addition, the company launched the Audio API, with six preset voices. Priced at $0.015 per 1,000 input characters, this API opens up new possibilities for text-to-speech applications.
Copyright Shield Program
An important step taken by OpenAI is the introduction of the Copyright Shield program. This initiative aims to protect companies that use OpenAI products against copyright claims. The company promises to cover legal costs if customers using "generally available" OpenAI developer platforms and ChatGPT Enterprise face IP lawsuits related to content created with OpenAI tools.
Other noteworthy announcements
Among other notable updates, OpenAI will offer companies a program to build their customer models in collaboration with researchers from the AI company. Removing the model selector in ChatGPT simplifies the user experience, and the launch of Whisper large v3, the next version of the open-source speech recognition model, adds to the diverse offering.
In addition, the company is doubling the number of tokens per limit rate for all paying GPT-4 customers, thereby increasing the value for users.