New GPT-4o Voice Mode features are rolling out

During the demo, GPT-4o instructed the demonstrator, Mark, on how to breathe better, including taking audio samples of his breath and giving him advice on how to do better. OpenAI CEO Sam Altman posted that the model is “natively multimodal,” which means the model could generate content or understand commands in voice, text, or images. Developers who want to tinker with GPT-4o will have access to the API, which is half the price and twice as fast as GPT-4 Turbo, Altman added on X. More and more tech companies and search engines are utilizing the chatbot to automate text or quickly answer user questions/concerns. OpenAI is forming a Collective Alignment team of researchers and engineers to create a system for collecting and “encoding” public input on its models’ behaviors into OpenAI products and services. This comes as a part of OpenAI’s public program to award grants to fund experiments in setting up a “democratic process” for determining the rules AI systems follow.

• It took words in Italian from Mira Murati and converted it to English, then took replies in English and translated to Italian.
• While this may have been little more than a typo-headline put live by mistake — meant to be for GPT-4-Turbo, it does add to the evidence a new version is coming.
• The hype is real and there are nearly 40,000 people watching the live stream on YouTube — so hopefully we get something interesting.
• ChatGPT is an AI chatbot with advanced natural language processing (NLP) that allows you to have human-like conversations to complete various tasks.

Enterprises will need to ensure they are using the model appropriately and implementing safeguards to prevent errors or misuse. These scores surpass those of highly regarded models like OpenAI’s GPT-4o and Anthropic’s Claude 3.5 Sonnet, catapulting Nvidia to the forefront of AI language understanding and generation. In the same time frame, OpenAI’s CEO Sam Altman shared a mysterious post on X that many suggest represents Orion. Some even asked ChatGPT to shed some light on it, and it looks like even the platform hinted at the next AI model. March 1, 2023 – OpenAI introduced the ChatGPT API for developers to integrate ChatGPT-functionality in their applications.

The new GPT-4o model

The release of Llama 3.1 marks a significant step forward for open-source LLMs. Whether it can truly compete with closed-source giants like GPT-4 remains to be seen. However, Meta’s efforts to democratise access to powerful AI tools and build a robust open-source ecosystem are likely to stimulate further innovation and exploration within the field. In the future, the model and its vision analysis capabilities will be expanded upon and added to consumer apps like ChatGPT, making its understanding of image and video more efficient. Most of the focus for GPT-4-Turbo is on improving the life of developers accessing the OpenAI model through an API call. The company says the new update will streamline workflows and create more efficient apps.

OpenAI staff showcased conversations with the AI chatbots with almost no lag, and the chatbot was even able to pivot quickly when interrupted. Elsewhere, the GPT Store, OpenAI’s library of and creation tools for third-party chatbots built on its AI models, is now available to users of ChatGPT’s free tier. Even though OpenAI released GPT-4 mere months after ChatGPT, we know that it took over two years to train, develop, and test. If GPT-5 follows a similar schedule, we may have to wait until late 2024 or early 2025. OpenAI has reportedly demoed early versions of GPT-5 to select enterprise users, indicating a mid-2024 release date for the new language model. The testers reportedly found that ChatGPT-5 delivered higher-quality responses than its predecessor.

Free users are getting a big upgrade

However, the model is still in its training stage and will have to undergo safety testing before it can reach end-users. For context, OpenAI announced the GPT-4 language model after just a few months of ChatGPT’s release in late 2022. GPT-4 was the most significant updates to the chatbot as it introduced a host of new features and under-the-hood improvements. Up until that point, ChatGPT relied on the older GPT-3.5 language model. For context, GPT-3 debuted in 2020 and OpenAI had simply fine-tuned it for conversation in the time leading up to ChatGPT’s launch. Generative Pre-trained Transformer, the model ChatGPT is based on, finds patterns within data sequences.

The temporary prototype is currently only available to a small group of users and its publisher partners, like The Atlantic, for testing and feedback. But the feature falls short as an effective replacement for virtual assistants. Volkswagen is taking its ChatGPT voice assistant experiment to vehicles in the United States. Its ChatGPT-integrated Plus Speech voice assistant is an AI chatbot based on Cerence’s Chat Pro product and a LLM from OpenAI and will begin rolling out on September 6 with the 2025 Jetta and Jetta GLI models. OpenAI announced it’s rolling out a feature that allows users to search through their ChatGPT chat histories on the web. The new feature will let users bring up an old chat to remember something or pick back up a chat right where it was left off.

On top of that, iOS 18 could see new AI-driven capabilities like being able to transcribe and summarize voice recordings. OpenAI has released an update to its advanced GPT-4-Turbo artificial intelligence model, bringing with it “majorly improved” responses and analysis capabilities. ChatGPT Plus users will get access to the app first, starting today, and a Windows version will arrive later in the year.

On the The TED AI Show podcast, former OpenAI board member Helen Toner revealed that the board did not know about ChatGPT until its launch in November 2022. Toner also said that Sam Altman gave the board inaccurate information about the safety processes the company had in place and that he didn’t disclose his involvement in the OpenAI Startup Fund. In a letter, Rep. Nancy Mace said Johansson’s testimony could “provide a platform” for concerns around deepfakes. OpenAI has banned a cluster of ChatGPT accounts linked to an Iranian influence operation that was generating content about the U.S. presidential election.

After the 90 days, the committee will share its safety recommendations with the OpenAI board, after which the company will publicly release its new security protocol. ChatGPT (and AI tools in general) have generated significant controversy for their potential implications for customer privacy and corporate safety. Additionally, Business Insider published a report about the release of GPT-5 around the same time as Altman’s interview with Lex Fridman. Sources told Business Insider that GPT-5 would be released during the summer of 2024. Altman could have been referring to GPT-4o, which was released a couple of months later. Therefore, it’s not unreasonable to expect GPT-5 to be released just months after GPT-4o.

According to the publication, it’s unclear if this version of Strawberry will provide any performance boosts to ChatGPT or GPT-4 this year. So, if you just want to try GPT-4o for a bit and are OK with waiting for the newest features, chat gpt 4 release then you probably don’t need a subscription to ChatGPT Plus. On the other hand, if you want to use GPT-4o often and have fun experimenting with the latest AI tools, you may consider the $20-a-month to be money well spent.

Led by previous investor Thrive Capital, the new cash brings OpenAI’s total raised to $17.9 billion, per Crunchbase. OpenAI denied reports that it is intending to release an AI model, code-named Orion, by December of this year. An OpenAI spokesperson told TechCrunch that they “don’t have plans to release a model code-named Orion this year,” but that leaves OpenAI substantial wiggle room.

However, they differ in their training models, data sources, user experiences and how they store data. Free account users will notice the biggest change as GPT-4o is not only better than the 3.5 model previously available in ChatGPT but also a boost on GPT-4 itself. Users will also now be able to run code snippets, analyze images and text files and use custom GPT chatbots. Unlike Anthropic, OpenAI retrains ChatGPT on user interactions by default, but it’s possible to opt out. One option is to not save chat history, with the caveat that the inability to refer back to previous conversations can limit the model’s usefulness. Users can also submit a privacy request to ask OpenAI to stop training on their data without sacrificing chat history — OpenAI doesn’t exactly make this process transparent or user-friendly, though.

Its ability to translate high scores into practical, valuable solutions will ultimately determine its long-term impact on the industry and society at large. Nvidia’s latest model release signals just how fast the AI landscape is shifting. While the long-term impact of Llama-3.1-Nemotron-70B-Instruct remains uncertain, its release marks a clear inflection point in the competition to build the most advanced AI systems. Like any AI system, Llama-3.1-Nemotron-70B-Instruct is not immune to risks. Nvidia has cautioned that the model has not been tuned for specialized domains like math or legal reasoning, where accuracy is critical.

There are several actions that could trigger this block including submitting a certain word or phrase, a SQL command or malformed data. AI is skilled at tapping into vast realms of data and tailoring it to a specific purpose—making it a highly customizable tool for combating misinformation. This new model enters the realm of complex reasoning, with implications for physics, coding, and more. Chen asked the model to read a bedtime story “about robots and love,” quickly jumping in to demand a more dramatic voice. The model got progressively more theatrical until Murati demanded that it pivot quickly to a convincing robot voice (which it excelled at).

Anthropic and OpenAI remain tight-lipped about their models’ specific sizes, architectures and training data. Both Claude and ChatGPT are estimated to have hundreds of billions of parameters. A recent paper from Anthropic suggested that Claude 3 has at least 175 billion, and a report by research firm SemiAnalysis estimated that GPT-4 has more than 1 trillion. Both also use transformer-based architectures, enhanced with techniques such as reinforcement learning from human feedback.

GPT-4o is OpenAI’s latest, fastest, and most advanced flagship model. However, the “o” in the title stands for “omni”, referring to its multimodal capabilities, which allow the model to understand text, audio, image, and video inputs and output text, audio, and image outputs. ZDNET’s recommendations are based on many hours of testing, research, and comparison shopping. We gather data from the best available sources, including vendor and ChatGPT App retailer listings as well as other relevant and independent reviews sites. And we pore over customer reviews to find out what matters to real people who already own and use the products and services we’re assessing. Based on rumors and leaks, we’re expecting AI to be a huge part of WWDC — including the use of on-device and cloud-powered large language models (LLMs) to seriously improve the intelligence of your on-board assistant.

Former OpenAI Staffer Says the Company Is Breaking Copyright Law and Destroying the Internet

Just days after OpenAI released GPT-4o, researchers noticed that many Chinese tokens included inappropriate phrases related to pornography and gambling. Model developers might have included these problematic tokens due to inadequate data cleaning, potentially degrading the model’s comprehension and risking security breaches and hallucinations. Unfortunately, each type of evidence — self-reported benchmarks from model developers, crowdsourced human evaluations and unverified anecdotes — has its own limitations. For developers building LLM apps and users integrating generative AI into their workflows, deciding which model is the best fit might ultimately require experimenting with both over time and in various contexts. Some developers, for example, say that they switch back and forth between GPT-4 and GPT-4o depending on the task at hand. You can foun additiona information about ai customer service and artificial intelligence and NLP. GPT-4 is a large language model (LLM) primarily designed for text processing, meaning that it lacks built-in support for handling images, audio and video.

Meta is planning to launch Llama-3 in several different versions to be able to work with a variety of other applications, including Google Cloud. Meta announced that more basic versions of Llama-3 will be rolled out soon, ahead of the release of the most advanced version, which is expected next summer. Llama-3 will also be multimodal, which means it is capable of processing and generating text, images and video. Therefore, it will be capable of taking an image as input to provide a detailed description of the image content. Equally, it can automatically create a new image that matches the user’s prompt, or text description.

That kind of stability can be crucial for critical and widely used applications, where reliability might be a higher priority than having the lowest costs or the latest features​. OpenAI now describes GPT-4o as its flagship model, and its improved speed, lower costs and multimodal capabilities will be appealing to many users. At the time, in mid-2023, OpenAI announced that it had no ChatGPT intentions of training a successor to GPT-4. However, that changed by the end of 2023 following a long-drawn battle between CEO Sam Altman and the board over differences in opinion. Altman reportedly pushed for aggressive language model development, while the board had reservations about AI safety. The former eventually prevailed and the majority of the board opted to step down.

The generative AI tool can answer questions and assist you with composing text, code, and much more. It’s an updated version of its GPT-4 model, which is now more than a year old. The new large language model, trained on vast amounts of data from the internet, will be better at handling text, audio and images in real-time. ChatGPT and Gemini are largely responsible for the considerable buzz around GenAI, which uses data from machine learning models to answer questions and create images, text and videos.

And while it still doesn’t know about events post-2021, GPT-4 has broader general knowledge and knows a lot more about the world around us. OpenAI also said the model can handle up to 25,000 words of text, allowing you to cross-examine or analyze long documents. Nvidia’s approach to creating Llama-3.1-Nemotron-70B-Instruct involved refining Meta’s open-source Llama 3.1 model using advanced training techniques, including Reinforcement Learning from Human Feedback (RLHF). This method allows the AI to learn from human preferences, potentially leading to more natural and contextually appropriate responses. ChatGPT is a general-purpose chatbot that uses artificial intelligence to generate text after a user enters a prompt, developed by tech startup OpenAI.

At its “Spring Update” the company is expected to announce something “magic” but very little is known about what we might actually see. Speculation suggestions a voice assistant, which would require a new AI voice model from the ChatGPT maker. OpenAI recently published a model rule book and spec, among the suggested prompts are those offering up real information including phone numbers and email for politicians.

The demonstrators also used the desktop version of GPT-4o to examine some code they had. Not only could GPT-4o explain what the code does, but it could tell what would happen if you tweaked specific parts of it. And the demonstrators showed that GPT-4o can act as a live translator, listening to two people speaking different languages and telling one person what the other said in their native tongue. These updates “had a much stronger response than we expected,” Altman told Bill Gates in January. These proprietary datasets could cover specific areas that are relatively absent from the publicly available data taken from the internet. Specialized knowledge areas, specific complex scenarios, under-resourced languages, and long conversations are all examples of things that could be targeted by using appropriate proprietary data.

While there is no official confirmation on the release plan by OpenAI or Microsoft, an OpenAI executive suggests that the next-generation AI model is expected to be 100 times more powerful than its predecessor. Jan 10, 2024 – With the launch of the GPT Store, ChatGPT users could discover and use other people’s custom GPTs. On this day, OpenAI also introduced ChatGPT Team, a collaborative tool for the workspace. November 6, 2023 – OpenAI announced the arrival of custom GPTs, which enabled users to build their own custom GPT versions using specific skills, knowledge, etc. August 28, 2023 – OpenAI launched ChatGPT Enterprise, calling it “the most powerful version of ChatGPT yet.” Benefits included enterprise-level security and unlimited usage of GPT-4.

OpenAI says Advanced Voice Mode might not launch for all ChatGPT Plus customers until the fall, depending on whether it meets certain internal safety and reliability checks. OpenAI and TIME announced a multi-year strategic partnership that brings the magazine’s content, both modern and archival, to ChatGPT. As part of the deal, TIME will also gain access to OpenAI’s technology in order to develop new audience-based products. The company says GPT-4o mini, which is cheaper and faster than OpenAI’s current AI models, outperforms industry leading small AI models on reasoning tasks involving text and vision.

Apple warns investors its new products might never be as profitable as the iPhone

If you’re using the free version of ChatGPT, you’re about to get a boost. On Monday, OpenAI debuted a new flagship model of its underlying engine, called GPT-4o, along with key changes to its user interface. The ChatGPT upgrade “brings GPT-4-level intelligence to everything, including our free users,” said OpenAI’s Mira Murati. OpenAI, citing the risk of misuse, says that it plans to first launch support for GPT-4o’s new audio capabilities to “a small group of trusted partners” in the coming weeks. “I think it is our job to live a few years in the future and remember that the tools we have now are going to kind of suck, looking backward at them, and that’s how we make sure the future is better,” he added. This website is using a security service to protect itself from online attacks.

It will affect the way people work, learn, receive healthcare, communicate with the world and each other. It will make businesses and organisations more efficient and effective, more agile to change, and so more profitable. It features several improvements compared to its predecessor, Llama-2. It is a more capable model that will eventually come with 400 billion parameters compared to a maximum of 70 billion for its predecessor Llama-2.

Another potential reasoning for making GPT-4 Turbo available for free is the imminent arrival of the next generation large language model from OpenAI. It means that more casual Copilot users will have a better experience, suffer fewer hallucinations from the chatbot and get more up to date information. What is the company announcing one day ahead of Google’s presumably AI-heavy I/O 2024 event tomorrow? According to Sam Altman, what OpenAI has in store is not the rumored search engine or GPT-5 updates, but we’ll find out what it really is at 1PM ET / 10AM PT. Dave is a freelance tech journalist who has been writing about gadgets, apps and the web for more than two decades. Based out of Stockport, England, on TechRadar you’ll find him covering news, features and reviews, particularly for phones, tablets and wearables.

OpenAI identified five website fronts presenting as both progressive and conservative news outlets that used ChatGPT to draft several long-form articles, though it doesn’t seem that it reached much of an audience. OpenAI struck a content deal with Hearst, the newspaper and magazine publisher known for the San Francisco Chronicle, Esquire, Cosmopolitan, ELLE, and others. The partnership will allow OpenAI to surface stories from Hearst publications with citations and direct links.

This was part of what prompted a much-publicized battle between the OpenAI Board and Sam Altman later in 2023. Altman, who wanted to keep developing AI tools despite widespread safety concerns, eventually won that power struggle. AGI, or artificial general intelligence, is the concept of machine intelligence on par with human cognition. A robot with AGI would be able to undertake many tasks with abilities equal to or better than those of a human.

ChatGPT is the AI-powered chatbot that made GenAI the hot technology of 2023. According to OpenAI CEO Sam Altman, ChatGPT reached 1 million users within five days of its release on Nov. 30, 2022. GenAI is still rapidly evolving, and models don’t always return correct answers. Despite the common occurrence of AI hallucinations — wrong answers generated by AI — in both ChatGPT and Gemini, the tools are being adopted by businesses and consumers seeking to automate time-consuming tasks. Right now its main benefit is in bringing massive reasoning, processing and natural language capabilities to the free version of ChatGPT for the first time.

As part of the event, OpenAI released multiple videos demonstrating the intuitive voice response and output capabilities of the model. The release of GPT-4o feels like a seminal moment for the future of AI chatbots. This technology pushes past much of the awkward latencies that plagued early chatbots. It’s easy to imagine a version of Siri that is quite useful with GPT-4o. These real-time capabilities are likely thanks to Nvidia’s latest inference chips, which Murati was sure to call out before ending the presentation. Regardless, OpenAI reaffirmed its dominant position as the leader in AI innovation with Monday’s demo.

The researchers were asked by OpenAI, the maker of GPT-4, to not release their prompts to the public, though they say they will provide them upon request. An even older version, GPT-3.5, is available for free but has a smaller context window. It has been a year since GPT-4 launched and in that time we’ve seen Google Gemini Ultra and Anthropic Claude 3 Opus achieve GPT-4 capabilities. Even French startup Mistral has a new model on par with OpenAI’s best and the company didn’t exist in March 2023. Customer trust is vital if it wants users to actually use the AI tools it is putting into every product.

• So, if you just want to try GPT-4o for a bit and are OK with waiting for the newest features, then you probably don’t need a subscription to ChatGPT Plus.
• GPT-5 is also expected to show higher levels of fairness and inclusion in the content it generates due to additional efforts put in by OpenAI to reduce biases in the language model.
• ChatGPT (and AI tools in general) have generated significant controversy for their potential implications for customer privacy and corporate safety.
• The version of Strawberry destined to become part of a chatbot is a smaller, simplified version of the AI.

The company’s new free flagship “omnimodel” looks like a supercharged version of assistants like Siri or Alexa. Lev Craig covers AI and machine learning as the site editor for TechTarget Editorial’s Enterprise AI site. GPT-4o is also designed to be quicker and more computationally efficient than GPT-4 across the board, not just for multimodal queries. According to OpenAI, GPT-4o is twice as fast as the most recent version of GPT-4. Subsequently, Johansson said she had retained legal counsel and revealed that Altman had previously asked to use her voice in ChatGPT, a request she declined.

It’s expected that Strawberry could help OpenAI’s problem with obtaining enough real world data for its LLMs to consume. As a result, the company is reportedly using the large version of Strawberry to train the successor of GPT-4, codenamed Orion. It’s also believed that the AI could be used to also improve the firm’s agents. A new report from The Information claims that OpenAI plans to launch a new AI, codename Strawberry, as part of a chatbot this fall. The outlet suspects that Strawberry could possibly become a part of ChatGPT. Although OpenAI has long been in the driver’s seat of the AI race, competitors have caught up to, and in some cases surpassed, GPT-4, leaving all eyes on the company’s next-generation large language model (LLM).