OpenAI Resurrects Cheaper, Faster GPT-3.5 Turbo That Now Powers Snapchat, Shopify, and More
OpenAI has formally brought back its highly capable GPT‑3.5 Turbo API to developers, a significant step towards restoring the foundation of the original ChatGPT that delighted the world in 2022.
The revival is also intended to enhance AI integration in top platforms, providing developers with an optimized and cost-effective solution for creating advanced chatbot capabilities and smart assistants. GPT‑3.5 Turbo Returns OpenAI Resurrects Cheaper, Faster GPT-3.5 Turbo That Now Powers Snapchat,
The GPT-3.5 Turbo API is now open for integration with different apps and services. OpenAI assured that this version is much more affordable and stable compared to its previous versions, at a cost of only $0.002 per 1,000 tokens, 10 times less expensive than previous GPT-3.5 versions.
But it's not just about affordability. This refreshed API version is designed for more than chat-based applications. Developers can now use it to power innovative features beyond text conversations, signaling OpenAI's push to broaden AI use cases across industries. Major Apps Leveraging GPT‑3.5 Turbo API
With OpenAI potentially requiring developers to accomplish ID verification, we expect to see newer AI models in the coming weeks. Before that, several well-known brands were already incorporating GPT‑3.5 Turbo into their platforms: Snapchat (Snap Inc.): Snapchat+ subscribers can now enjoy "My AI," an individualized chatbot providing text edits, suggestions, and instant conversation augmentation.
Quizlet: Supporting more than 60 million students worldwide, the app will incorporate GPT‑3.5 Turbo to act as an interactive AI tutor, adjusting to users' levels of study and subjects.
Instacart: The "Ask Instacart" feature launching soon will enable shoppers to communicate with the platform in natural language queries—such as asking for recipe recommendations in relation to shopping lists.
Shopify: GPT-3.5 integration will enable an AI-based shopping assistant that provides personalized product suggestions to over 100 million users. Looking Back at GPT-3.5 and How It Came to Be
Released in November 2022, GPT‑3.5 became the basis of ChatGPT's free tier, remaining active until replaced by GPT-4o mini in mid-2024. Although it was widely used, GPT‑3.5 was criticized for producing intermittent "hallucinations" due to outdated training data (dating only through September 2021). The new Turbo variant removes these issues with increased stability and wider utility. Whisper API Gets a Major Upgrade
In addition to GPT‑3.5 Turbo's return, OpenAI has upgraded its Whisper API, the company's open-source speech-to-text model, Digital Trends reports.
Originally released in 2022, Whisper now offers faster processing and greater compatibility across audio formats like MP3, M4A, WAV, and WebM.
One of its most notable implementations is Speak, South Korea's leading English learning app. The Whisper API upgrade will support the app's global expansion and enhance its ability to offer open-ended, accurate language learning experiences. OpenAI's Strategic Shift on Open-Source and Older Models
In the face of increasing competition, particularly from Chinese AI brand DeepSeek, OpenAI is rethinking its open-source approach. CEO Sam Altman admitted recently in an AMA that the company had been "on the wrong side of history," suggesting greater openness and access for earlier models.
OpenAI's Chief Product Officer, Kevin Weil, also revealed the company might open-source more legacy models in the future. The Whisper API serves as a leading example of this shift.
Originally published on Tech Times
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Int'l Business Times
4 days ago
- Int'l Business Times
US Judge Backs Using Copyrighted Books To Train AI
A US federal judge has sided with Anthropic regarding training its artificial intelligence models on copyrighted books without authors' permission, a decision with the potential to set a major legal precedent in AI deployment. District Court Judge William Alsup ruled on Monday that the company's training of its Claude AI models with books bought or pirated was allowed under the "fair use" doctrine in the US Copyright Act. "Use of the books at issue to train Claude and its precursors was exceedingly transformative and was a fair use," Alsup wrote in his decision. "The technology at issue was among the most transformative many of us will see in our lifetimes," Alsup added in his 32-page decision, comparing AI training to how humans learn by reading books. Tremendous amounts of data are needed to train large language models powering generative AI. Musicians, book authors, visual artists and news publications have sued various AI companies that used their data without permission or payment. AI companies generally defend their practices by claiming fair use, arguing that training AI on large datasets fundamentally transforms the original content and is necessary for innovation. "We are pleased that the court recognized that using 'works to train LLMs was transformative,'" an Anthropic spokesperson said in response to an AFP query. The judge's decision is "consistent with copyright's purpose in enabling creativity and fostering scientific progress," the spokesperson added. The ruling stems from a class-action lawsuit filed by authors Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson, who accused Anthropic of illegally copying their books to train Claude, the company's AI chatbot that rivals ChatGPT. However, Alsup rejected Anthropic's bid for blanket protection, ruling that the company's practice of downloading millions of pirated books to build a permanent digital library was not justified by fair use protections. Along with downloading of books from websites offering pirated works, Anthropic bought copyrighted books, scanned the pages and stored them in digital format, according to court documents. Anthropic's aim was to amass a library of "all the books in the world" for training AI models on content as deemed fit, the judge said in his ruling. While training AI models on the pirated content posed no legal violation, downloading pirated copies to build a general-purpose library constituted copyright infringement, regardless of eventual training use. The case will now proceed to trial on damages related to the pirated library copies, with potential penalties including financial damages. Anthropic said it disagreed with going to trial on this part of the decision and was evaluating its legal options. Valued at $61.5 billion and heavily backed by Amazon, Anthropic was founded in 2021 by former OpenAI executives. The company, known for its Claude chatbot and AI models, positions itself as focused on AI safety and responsible development.
Int'l Business Times
4 days ago
- Int'l Business Times
Phoebe Gates Taps ChatGPT To Power Viral Growth Of Fashion Startup Phia
Phoebe Gates, daughter of tech giant Bill Gates, is carving out her own space in the fashion technology world. Her startup, Phia, is an AI-driven shopping assistant designed to help users find the best deals on clothing and accessories by scanning millions of listings across popular resale and retail platforms. What makes Phia stand out is not just its tech — it also uses ChatGPT to fuel creative marketing efforts. Gates and her co-founder Sophia Kianni revealed how they studied viral TikTok videos using AI to understand what makes content popular. This insight helped them craft engaging videos that boosted Phia's visibility rapidly, as reported by Business Insider. Phia aggregates listings from websites like eBay, Poshmark and The RealReal, offering users instant feedback on whether a price is a good deal with a simple "Should I Buy This?" button. This feature helps shoppers navigate the growing secondhand market with confidence, according to The Verge. Despite her famous last name, Phoebe insists she built Phia without financial help from her parents. On their podcast, ''The Burnouts,'' she and Kianni shared their focus on sustainability and smart shopping, highlighting their desire to create an app that supports eco-conscious consumers, reported People. By combining innovative AI tools with savvy social media strategies, Phoebe Gates and her team have positioned Phia as an exciting new player in online retail, especially among young shoppers who care about sustainability and value.
Int'l Business Times
19-06-2025
- Int'l Business Times
Justice At Stake As Generative AI Enters The Courtroom
Generative artificial intelligence (GenAI) is making its way into courts despite early stumbles, raising questions about how it will influence the legal system and justice itself. Judges use the technology for research, lawyers utilize it for appeals and parties involved in cases have relied on GenAI to help express themselves in court. "It's probably used more than people expect," said Daniel Linna, a professor at the Northwestern Pritzker School of Law, about GenAI in the US legal system. "Judges don't necessarily raise their hand and talk about this to a whole room of judges, but I have people who come to me afterward and say they are experimenting with it." In one prominent instance, GenAI enabled murder victim Chris Pelkey to address an Arizona courtroom -- in the form of a video avatar -- at the sentencing of the man convicted of shooting him dead in 2021 during a clash between motorists. "I believe in forgiveness," said a digital proxy of Pelkey created by his sister, Stacey Wales. The judge voiced appreciation for the avatar, saying it seemed authentic. "I knew it would be powerful," Wales told AFP, "that that it would humanize Chris in the eyes of the judge." The AI testimony, a first of its kind, ended the sentencing hearing at which Wales and other members of the slain man's family spoke about the impact of the loss. Since the hearing, examples of GenAI being used in US legal cases have multiplied. "It is a helpful tool and it is time-saving, as long as the accuracy is confirmed," said attorney Stephen Schwartz, who practices in the northeastern state of Maine. "Overall, it's a positive development in jurisprudence." Schwartz described using ChatGPT as well as GenAI legal assistants, such as LexisNexis Protege and CoCounsel from Thomson Reuters, for researching case law and other tasks. "You can't completely rely on it," Schwartz cautioned, recommending that cases proffered by GenAI be read to ensure accuracy. "We are all aware of a horror story where AI comes up with mixed-up case things." The technology has been the culprit behind false legal citations, far-fetched case precedents, and flat-out fabrications. In early May, a federal judge in Los Angeles imposed $31,100 in fines and damages on two law firms for an error-riddled petition drafted with the help of GenAI, blasting it as a "collective debacle." The tech is also being relied on by some who skip lawyers and represent themselves in court, often causing legal errors. And as GenAI makes it easier and cheaper to draft legal complaints, courts already overburdened by caseloads could see them climb higher, said Shay Cleary of the National Center for State Courts. "Courts need to be prepared to handle that," Cleary said. Law professor Linna sees the potential for GenAI to be part of the solution though, giving more people the ability to seek justice in courts made more efficient. "We have a huge number of people who don't have access to legal services," Linna said. "These tools can be transformative; of course we need to be thoughtful about how we integrate them." Federal judges in the US capitol have written decisions noting their use of ChatGPT in laying out their opinions. "Judges need to be technologically up-to-date and trained in AI," Linna said. GenAI assistants already have the potential to influence the outcome of cases the same way a human law clerk might, reasoned the professor. Facts or case law pointed out by GenAI might sway a judge's decision, and could be different than what a legal clerk would have come up with. But if GenAI lives up to its potential and excels at finding the best information for judges to consider, that could make for well-grounded rulings less likely to be overturned on appeal, according to Linna.
