Abu Zar Mishwani
During DevDay, an event highlighting the latest advancements in artificial intelligence, OpenAI introduced a range of new features and enhancements for developers and ChatGPT users. One notable addition is the ability to create custom GPTs.
Let’s first take a closer look at the most recent version of GPT…
GPT-4 Turbo
One of the noteworthy highlights from the event was the announcement of GPT-4 Turbo, an enhanced version of their already powerful AI model. This new version aims to deliver exceptional efficiency and a stronger grasp of current global developments.
GPT-4 Turbo comes with extended features, such as awareness of global happenings until April 2023 and an impressive 128k context window. This means it can handle over 300 pages of text in a single request.
These improvements also come with a significant price reduction, making it three times more affordable for input tokens and twice as cost-effective for output tokens when compared to its predecessor, GPT-4.
The newly introduced model, currently available in a preview phase, is open to all paying developers. OpenAI has additionally made enhancements to function calling, enabling more precise and intricate multi-function instructions within a single message.
The introduction of JSON mode for developers is another notable enhancement, ensuring that the model produces valid JSON outputs. This feature is especially valuable for tasks that require generating specific formats.
Assistants API
OpenAI has unveiled the Assistants API with the goal of streamlining the development of AI-powered applications. It offers features like a Code Interpreter, Retrieval, and enhanced function calling to facilitate the process.
The introduction of this new API represents a stride towards more sophisticated and adaptable AI tools, enabling them to tackle intricate tasks with greater efficiency.
Updates To GPT-3.5 Turbo
In addition, OpenAI has made improvements to its GPT-3.5 Turbo model, which now comes with a default 16K context window and enhanced performance across a range of tasks.
These updates showcase a notable 38% enhancement in handling format-related tasks, including the generation of JSON, XML, and YAML.
Multimodality
In the field of visual processing, GPT-4 Turbo has gained the ability to analyze images, a capability that companies like BeMyEyes are already using to provide assistance to individuals with visual impairments.
Moreover, developers now have the option to incorporate DALL·E 3 into their applications for image generation. OpenAI has also implemented a strong moderation system to safeguard against misuse.
OpenAI has not neglected the importance of human-like interactions and has introduced a text-to-speech API that delivers human-like speech generation. This API offers a variety of voice presets and model variants suitable for a wide range of applications.
GPT-4 Fine Tuning
The company is also granting experimental access to GPT-4 fine-tuning, focusing on enhancing both quality and safety aspects.
To cater to organizations in need of tailored solutions, OpenAI has initiated the Custom Models program. This program provides personalized GPT-4 training and exclusive access to custom models for specific needs.
Price Drops For Developers
OpenAI has also revealed price reductions for a range of services, along with expanded rate limits to facilitate the scaling of applications.
One of the most significant changes is the pricing, with GPT-4 Turbo input tokens now costing three times less than GPT-4 at just $0.01 and output tokens being two times cheaper at $0.03.
Copyright Shield
OpenAI has taken steps to bolster its commitment to copyright protection by introducing Copyright Shield. This measure is designed to safeguard ChatGPT enterprise users and API developers from potential copyright issues.
Speech Recognition
Furthermore, OpenAI is elevating its speech recognition capabilities through the introduction of Whisper large-v3 and the Consistency Decoder, aiming to deliver improved performance and enhanced image quality.
Custom GPTs
OpenAI has introduced GPTs, a feature that empowers users to customize ChatGPT for particular purposes, promoting community involvement in the advancement of AI technology.
This feature will be complemented by the forthcoming GPT Store, which will serve as a platform to showcase GPTs created by users.
GPT-4 All Tools
OpenAI has simplified the usage of ChatGPT, powered by GPT-4 Turbo. This version encompasses information up to April 2023 and eliminates the necessity to switch between models, making the user experience more convenient.
Conclusion
The recent breakthroughs presented by OpenAI at DevDay mark a significant stride in AI technology, providing developers and businesses with more potent, efficient, and easily accessible tools.
These innovations not only reflect the ever-evolving landscape of AI but also highlight the commitment to making AI more accessible and impactful across various sectors, from communication to problem-solving.
The future of AI indeed looks promising, with these advancements paving the way for exciting possibilities and transformative applications.