It’s been an exciting journey observing OpenAIβs progressive leaps in the field of generative AI. Their cutting-edge models, gpt-3.5-turbo
and gpt-4
, released earlier this year, have been at the forefront of developersβ toolkits, catalyzing countless fascinating applications.
We have discussed many GPT applications on the Finxter blog, such as productivity use cases for coders and browser-based AutoGPT tools, but today I’ll show you the most recent model updates that will redefine the game once again. π€―
New Function Calling Capability: Transcending Language Barriers
The most transformative of these updates is the introduction of a new function calling capability in their Chat Completions API
for the gpt-4-0613
and gpt-3.5-turbo-0613
models. This might sound a bit technical, but the implications are huge.
Developers can now describe functions to these models, and, based on the user’s input, the models will choose to output a JSON object that contains arguments to call those functions. This is like a new language between GPT’s capabilities and external tools and APIs.
This opens up a new realm of possibilities. Imagine asking a chatbot, βWhatβs the weather like in Boston?β
and it translates your question into a function call that retrieves real-time weather data.
π€οΈ Recommended: Building a Weather App with Python
Or, if youβre a business owner, you could ask, βWho are my top ten customers this month?β
and your chatbot could convert that into an internal API call to pull the data from your database. This upgrade aims to deliver structured data more reliably, enhancing the user experience and expanding the realm of possibilities for AI-driven applications.
Enhanced Models: GPT-4 and GPT-3.5 Turbo
The AI squad has also supercharged the existing gpt-4
and gpt-3.5-turbo
models. The gpt-4-0613
model now comes equipped with the same function calling capabilities, and the gpt-4-32k-0613
version offers an extended context length, allowing it to understand and handle larger text bodies more effectively.
In the same vein, the gpt-3.5-turbo-0613
includes function calling and better steerability. The gpt-3.5-turbo-16k
version offers an impressive four times the context length of its predecessor, supporting nearly 20 pages of text in a single request.
Phasing Out Old Models: Upgrades and Deprecations
In line with their forward-thinking approach, OpenAI has started the process of upgrading and retiring the initial versions of gpt-4
and gpt-3.5-turbo
.
Applications using the stable model names will automatically be updated to the new models on June 27th. However, developers who need more time to adapt can continue using the older models until September 13th.
More Bang for Your Buck: Significant Price Reductions
OpenAI has gone above and beyond to ensure these revolutionary tools remain accessible. With efficiency enhancements, they’re cutting prices significantly.
The cost of using their popular embeddings model has been reduced by a whopping 75%, and the input tokens for the widely used gpt-3.5-turbo
model are now 25% cheaper.
These updates reflect OpenAI’s continuous commitment to pushing the boundaries of artificial intelligence at an exponentially decreasing cost.
π©βπ» I, for one, am incredibly excited to see how these advancements will be put to use by the global developer community. This wave of updates not only enriches the capabilities of these models but also sets a new standard for AI’s interaction with external tools and APIs.
Potential use cases of this new function calling capability are manifold:
In the realm of customer service, we could witness chatbots providing real-time responses by connecting with external databases, APIs, or tools, thereby significantly improving customer interactions.
For businesses relying on internal data systems, transforming natural language queries into API calls or database queries could streamline the process of data retrieval and manipulation.
Meanwhile, educators and researchers who often deal with larger texts will benefit from the expanded context length offered by the gpt-4-32k-0613
and gpt-3.5-turbo-16k
models. π€©
This feature could enhance text comprehension and analysis in contexts ranging from comprehensive literature reviews to extensive legal documents.
OpenAI Glossary Cheat Sheet (100% Free PDF Download) π
Finally, check out our free cheat sheet on OpenAI terminology, many Finxters have told me they love it! β₯οΈ
π‘ Recommended: OpenAI Terminology Cheat Sheet (Free Download PDF)