OpenAI's long-awaited GPT-5 model nears release
Two early testers of the new model told Reuters they have been impressed with its ability to code and solve science and math problems, but they believe the leap from GPT-4 to GPT-5 is not as large as the one from GPT-3 to GPT-4. The testers, who have signed non-disclosure agreements, declined to be named for this story.
OpenAI declined to comment for this story.
GPT-4's leap was based on more compute power and data, and the company was hoping that 'scaling up' in a similar way wouldconsistently lead to improved AI models.
But OpenAI, which is backed by Microsoft and is currently valued at $300 billion, ran into issues scaling up. One problem was the data wall the company ran into, and OpenAI's former chief scientist Ilya Sutskever said last year that while processing power was growing, the amount of data was not.
He was referring to the fact that large language models are trained on massive datasets that scrape the entire internet, and AI labs have no other options for large troves of human-generated textual data.
Apart from the lack of data, another problem was that 'training runs' for large models are more likely to have hardware-induced failures given how complicated the system is, and researchers may not know the eventual performance of the models until the end of the run, which can take months.
OpenAI has not said when GPT-5 will be released, but the industry expects it to be any day now, according to media reports. Boris Power, head of Applied Research at OpenAI, said in an X post on Monday: "Excited to see how the public receives GPT-5."
'OpenAI made such a great leap from GPT-3 to GPT-4, that ever since then, there has been an enormous amount of anticipation over GPT-5,' said Navin Chaddha, managing partner at venture capital fund Mayfield, who invests in AI companies but is not an OpenAI investor. 'The hope is that GPT-5 will unlock AI applications that move beyond chat into fully autonomous task execution."
'TEST-TIME COMPUTE'
Nearly three years ago, ChatGPT introduced the world to generative AI, dazzling users with its ability to write humanlike prose and poetry, quickly becoming one of the fastest growing apps ever.
In March 2023, OpenAI followed up ChatGPT with the release of GPT-4, a large language model that made huge leaps forward in intelligence. While GPT-3.5, an earlier version of the model, received a bar exam score in the bottom 10%, GPT-4 passed the simulated bar exam in the top 10%.
GPT-4 then became the model to beat and the world came to terms with the fact thatAI models could outperform humans in many tasks.
Soon, other companies were catching on. The same year, Alphabet's Google and Anthropic - which is backed by Amazon and Google - released competitive models to GPT-4. Within a year,open-source models on par with GPT-4 such as Meta Platforms' Llama 3 models were released.
Along with training large models, OpenAI has now invested in another route, called 'test-time compute,' which channels more processing power to solve challenging tasks such as math or complex operations that demand human-like reasoning and decision-making.
The company's CEO Sam Altman said earlier this year that GPT-5 would combine both test-time compute and its large models. He also said that OpenAI's model and product offerings had become "complicated."
(Reporting by Anna Tong in San Francisco; Editing by Sayantani Ghosh and Muralikumar Anantharaman)
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
The Star
8 minutes ago
- The Star
Malaysia's semiconductor exports still exempted from reciprocal tariffs
FILE PHOTO: Semiconductor chips are seen on a circuit board of a computer in this illustration picture taken February 25, 2022. REUTERS/Florence Lo/Illustration/File Photo KUALA LUMPUR: The Ministry of Investment, Trade and Industry (MITI) reiterated that Malaysia's semiconductor exports to the United States (US) are currently still exempted from reciprocal tariffs. However, the ministry said that this is subject to the US's current policy reviews and changes at any time. MITI minister Tengku Datuk Seri Zafrul Abdul Aziz said the US is currently conducting an investigation under Section 232 of its Trade Expansion Act of 1962 to determine whether imports of semiconductors, semiconductor manufacturing equipment and related derivative products pose a threat to its national security. "The investigation was supposed to be completed by the end of December, but (has been) expedited, and the results from this investigation will be used for further action, including the possible measure of high tariffs as announced by US President Donald Trump this morning,' he said during the Minister's Question Time session in the Dewan Rakyat today. He was replying to a supplementary question from Lim Guan Eng (Bagan-PH) about the impact of the latest reports that the US would impose a tariff of around 100 per cent on imported semiconductors from countries that do not produce them in the US or do not plan to do so. Tengku Zafrul said Malaysia is currently taking several mitigation measures, including contacting directly with the US Trade Representative and the Department of Commerce to obtain official clarification and to advocate for the interests of Malaysian exporters. He said that if tariffs are imposed, Malaysia will risk losing a key market in the US, as locally produced products might become less competitive. However, he said, as of now, the US government has not finalised the detailed mechanism for implementing the tariffs. "Based on current information, the exemption is not country-based but on companies that invest directly in the US, regardless of the company's location in another country. "Therefore, the impact is highly dependent on the investments and operations of international companies located in Malaysia,' he added. Tengku Zafrul also stated that 65 per cent of Malaysia's semiconductor exports to the US are from American companies operating in Malaysia. - Bernama
The Star
37 minutes ago
- The Star
OpenAI releases free, downloadable models in competition catch-up
SAN FRANCISCO: OpenAI on Tuesday released two new artificial intelligence (AI) models that can be downloaded for free and altered by users, to challenge similar offerings by US and Chinese competition. The release of gpt-oss-120b and gpt-oss-20b "open-weight language models" comes as the ChatGPT-maker is under pressure to share inner workings of its software in the spirit of its origin as a nonprofit. "Going back to when we started in 2015, OpenAI's mission is to ensure AGI (Artificial General Intelligence) that benefits all of humanity," said OpenAI chief executive Sam Altman. An open-weight model, in the context of generative AI, is one in which the trained parameters are made public, enabling users to fine-tune it. Meta touts its open-source approach to AI, and Chinese AI startup DeepSeek rattled the industry with its low-cost, high-performance model boasting an open weight approach that allows users to customise the technology. "This is the first time that we're releasing an open-weight model in language in a long time, and it's really incredible," OpenAI co-founder and president Greg Brockman said during a briefing with journalists. The new, text-only models deliver strong performance at low cost, according to OpenAI, which said they are suited for AI jobs like searching the internet or executing computer code, and are designed to be easy to run on local computer systems. "We are quite hopeful that this release will enable new kinds of research and the creation of new kinds of products," Altman said. OpenAI said it is working with partners including French telecommunications giant Orange and cloud-based data platform Snowflake on real-world uses of the models. The open-weight models have been tuned to thwart being used for malicious purposes, according to OpenAI. Altman early this year said his company had been "on the wrong side of history" when it came to being open about how its technology works. He later announced that OpenAI will continue to be run as a nonprofit, abandoning a contested plan to convert into a for-profit organisation. The structural issue had become a point of contention, with major investors pushing for better returns. That plan faced strong criticism from AI safety activists and co-founder Elon Musk, who sued the company he left in 2018, claiming the proposal violated its founding philosophy. In the revised plan, OpenAI's money-making arm will be open to generate profits but will remain under the nonprofit board's supervision. – AFP
The Star
37 minutes ago
- The Star
Exclusive-Rubio orders US diplomats to launch lobbying blitz against Europe's tech law
WASHINGTON (Reuters) -President Donald Trump's administration has instructed U.S. diplomats in Europe to launch a lobbying campaign to build an opposition to the European Union's Digital Services Act, which Washington says stifles free speech and imposes costs on U.S. tech companies, an internal diplomatic cable seen by Reuters showed. In a State Department cable dated August 4 that was signed by U.S. Secretary of State Marco Rubio, the agency said the EU was pursuing "undue" restrictions on freedom of expression by its efforts to combat hateful speech, misinformation and disinformation and the DSA was further enhancing these curbs. The EU's DSA is a landmark law that is meant to make the online environment safer and fairer by compelling tech giants to do more to tackle illegal content, including hate speech and child sexual abuse material. Trump has made combating censorship - particularly what he sees as the stifling of conservative voices online - a major theme of his administration. Top U.S. officials, including Vice President JD Vance, have focused on European officials and regulations, accusing them of "censoring" Americans, an accusation that the European Union rejects. The cable, whose headline described it as an "action request", tasked American diplomats across U.S. embassies in Europe with regularly engaging with EU governments and digital services authorities to convey U.S. concerns about the DSA and the financial costs for U.S. tech companies. "Posts should focus efforts to build host government and other stakeholder support to repeal and/or amend the DSA or related EU or national laws restricting expression online," the cable said in its "objective" section, referring to U.S. diplomatic missions. It provided specific suggestions to U.S. diplomats on how the EU law may be changed and the talking points to help them make that argument. State Department did not comment for this story. EU tech chief Henna Virkkunen's office did not immediately respond to a request for comment. In March, EU's antitrust and tech chiefs told U.S. lawmakers that the new tech rule aimed to keep digital markets open and is not targeted at U.S. tech giants. The Commission has also pushed back against speculation that the 27-member EU's landmark tech regulatory regime could be included in the EU-U.S. negotiations. "Our legislation will not be changed. The DMA and the DSA are not on the table in the trade negotiations with the U.S.," Commission spokesperson Thomas Regnier told a daily news conference. A CAMPAIGN FOR "FREE SPEECH" The order to U.S. diplomats marks an acceleration of the administration's efforts to promote what it calls "America's free-speech tradition," a policy that has added friction to the already fraught U.S. relationship with European allies. That policy came into focus in February, when Vance stunned European leaders by accusing them - at a conference usually known for displays of transatlantic unity - of censoring the speech of groups such as Germany's right-wing AfD party and backsliding on democracy. During his trip, Vance went on to meet with the leaders of AfD — classified by Germany's domestic intelligence service as a suspected extremist group — which became the country's largest opposition party after the February election. Trump and his Republican allies have repeatedly accused the administration of Democratic former President Joe Biden of encouraging suppression of free speech on online platforms, claims that have centered on efforts to stem false claims about vaccines and elections. The U.S. Supreme Court ruled last year that the Biden administration's contacts with social media companies did not violate America's First Amendment protections around free speech. The directive by the State Department ordered U.S. diplomats to investigate any claims of censorship which it described as "any government efforts to suppress protected forms of expression or coerce private companies to do the same", adding that the priority should be given to any incidents that impact U.S. citizens and companies. Examples could include arrests, court cases, property seizures and online suspensions, it said. "Posts should meet with government officials, businesses, civil society, and impacted individuals to report on censorship cases, including but not limited to those related to the DSA," the cable said. In March, the chairman of the U.S. Federal Communications Commission (FCC) specifically criticized DSA saying it was not compatible with America's free speech tradition. In May, Rubio threatened visa bans for people who "censor" speech by Americans, including on social media, and suggested the policy could target foreign officials regulating U.S. tech companies. "OVERLY BROAD CONTROLS" U.S. tech companies like Facebook and Instagram parent Meta have weighed in too, saying the DSA amounts to censorship of their platforms. Tesla Chief Executive Elon Musk, who also owns social media company X, was a leading adviser to the U.S. president before the two fell out, while the bosses of Amazon, Meta and Google-owner Alphabet (GOOGL.O) took prominent spots at Trump's inauguration in January. Rubio's directive takes particular aim at DSA's description of illegal content, saying it was expansive and told U.S. diplomats to advocate to get the definition of "illegal content" narrowed so that it would not curb freedom of expression, including in political and religious discourse. Another suggestion was to withdraw or amend the Code of Conduct on Disinformation, a framework under DSA, which the State Department said was setting "overly broad controls" on content in a way that was undermining freedom of expression. Other talking points included removing or reducing fines for non-compliance to content restrictions and not relying on "trusted flaggers", entities designated by national authorities to report illegal online content to platforms. (Reporting by Humeyra Pamuk; editing by Diane Craft)
