5, described as its 'smartest and most intuitive model yet', according to multiple reports. 4 on coding tests. 5 is being rolled out to Plus, Pro, Business, and Enterprise users.
It includes 'strongest safeguards to date' and was tested by nearly 200 early-access partners. 5 achieves similar advanced cybersecurity performance to Anthropic's Mythos Preview. 6%, within margin of error.
We believe that advanced AI systems can help researchers move through these workflows faster–not just by making existing work more efficient, but by helping scientists explore more possibilities, surface connections that might otherwise be missed, and arrive at better hypotheses sooner.
The model succeeded in some advanced tasks that previous models could not handle at all. 5 is designed to carry out tasks like online research and data analysis without human oversight. 5.
4 Cyber, a variant focused on cyber defence with fewer restrictions on cybersecurity queries, according to multiple reports. 4 Cyber includes capabilities for binary reverse engineering. Access is limited to vetted security vendors, organisations and researchers through the Trusted Access for Cyber programme.
We view it as the beginning of a long-term commitment to building AI that can accelerate scientific discovery in areas that matter deeply to society, from human health to broader biological research.
OpenAI also launched GPT-Rosalind, a new AI model for biology, drug discovery, and translational medicine, named after Rosalind Franklin. Meanwhile, Anthropic released Claude Mythos Preview, a model that can identify thousands of previously unknown, high-severity vulnerabilities, but is too dangerous for full public release according to Anthropic. OpenAI has updated its mission statement with five guiding principles for AGI, according to multiple reports.
Sam Altman wrote that holding back models that might pose a threat was a 'misplaced worry'. , 3:1 and 1:3), and more languages, according to multiple reports. It is available to all ChatGPT and Codex users, with advanced features requiring a subscription.
Power in the future can either be held by a small handful of companies using and controlling superintelligence, or it can be held in a decentralised way by people.
5 Instant will stop using unjustified emojis in its responses. 5 Instant. 5 Instant can be used for free until May 6.
We believe the latter is much better, and our goal is to put truly general AI in the hands of as many people as possible... Our mission is to ensure that AGI benefits all of humanity.
This allows us to expand access based on evidence and real signals of trust, rather than relying on manual decisions.
We don’t think it’s practical or appropriate to centrally decide who gets to defend themselves. Instead, we aim to enable as many legitimate defenders as possible, with access grounded in verification, trust signals, and accountability.