OpenAI waxay soo saartay GPT-6 Astra, waxayna sheegtay inay ku jirto xilli AGI-ga
OpenAI ayaa maanta soo saartay nidaamkeeda AI-ka cusub ee GPT-6 Astra, waxayna sheegtay inuu yahay midka ugu horreeya ee gaaray xadka amniga cyber-ka ee shirkaddu u qoondeysay, iyadoo madaxweynaha shirkaddu uu sheegay inay ku jirto xilli AGI-ga.
GPT-6 Astra: Midka ugu horreeya ee gaaray xadka amniga cyber-ka
OpenAI ayaa soo saartay GPT-6 Astra, kaas oo ah midka ugu horreeya ee gaaray xadka amniga cyber-ka ee shirkaddu u qoondeysay, iyadoo shirkaddu ay sheegtay inuu yahay midka ugu horreeya ee gaaray xadka amniga cyber-ka ee shirkaddu u qoondeysay. Nidaamkan cusub ayaa la soo bandhigay macaamiisha enterprise-ka ee amniga cyber-ka, waxaana la filayaa inuu dib u soo baxo macaamiisha kale ee Plus, Pro, Business, iyo Enterprise muddo kooban gudaheed. GPT-6 Astra ayaa sidoo kale la heli doonaa API-ga OpenAI iyo AWS.
Shirkaddu waxay sheegtay in GPT-6 Astra uu yahay midka ugu horreeya ee gaaray xadka amniga cyber-ka ee shirkaddu u qoondeysay, waxaana la sheegay inuu yahay midka ugu horreeya ee gaaray xadka amniga cyber-ka ee shirkaddu u qoondeysay. Nidaamkan cusub ayaa la soo bandhigay macaamiisha enterprise-ka ee amniga cyber-ka, waxaana la filayaa inuu dib u soo baxo macaamiisha kale ee Plus, Pro, Business, iyo Enterprise muddo kooban gudaheed.
Greg Brockman: Waxaan ku jirnaa xilli AGI-ga
Greg Brockman, madaxweynaha OpenAI, wuxuu sheegay in GPT-6 Astra uu noqon karo midka abuuray AGI-ga, waxaana uu yiri, 'Haddii aan ku dhawaado sanooyin, waxaan u arkaa inay noqon karto waqtigan.' Wuxuu sidoo kale sheegay in shirkaddu ay ku jirto xilli AGI-ga, waxaana uu sheegay in GPT-6 Astra uu yahay midka ugu horreeya ee gaaray xadka amniga cyber-ka ee shirkaddu u qoondeysay.
Greg Brockman wuxuu sheegay in GPT-6 Astra uu noqon karo midka abuuray AGI-ga, waxaana uu yiri, 'Haddii aan ku dhawaado sanooyin, waxaan u arkaa inay noqon karto waqtigan.' Wuxuu sidoo kale sheegay in shirkaddu ay ku jirto xilli AGI-ga, waxaana uu sheegay in GPT-6 Astra uu yahay midka ugu horreeya ee gaaray xadka amniga cyber-ka ee shirkaddu u qoondeysay.
Amniga iyo isku-dhafka: OpenAI waxay isku dayaysaa inay dib u soo celiso magaca
OpenAI waxay sheegtay in GPT-6 Astra uu yahay midka ugu horreeya ee gaaray xadka amniga cyber-ka ee shirkaddu u qoondeysay, waxaana la sheegay inuu yahay midka ugu horreeya ee gaaray xadka amniga cyber-ka ee shirkaddu u qoondeysay. Shirkaddu waxay sheegtay inay ku jirto xilli AGI-ga, waxaana uu sheegay in GPT-6 Astra uu yahay midka ugu horreeya ee gaaray xadka amniga cyber-ka ee shirkaddu u qoondeysay.
OpenAI waxay sheegtay in GPT-6 Astra uu yahay midka ugu horreeya ee gaaray xadka amniga cyber-ka ee shirkaddu u qoondeysay, waxaana la sheegay inuu yahay midka ugu horreeya ee gaaray xadka amniga cyber-ka ee shirkaddu u qoondeysay. Shirkaddu waxay sheegtay inay ku jirto xilli AGI-ga, waxaana uu sheegay in GPT-6 Astra uu yahay midka ugu horreeya ee gaaray xadka amniga cyber-ka ee shirkaddu u qoondeysay.
OpenAI has unveiled GPT-6 Astra, a new AI model it says marks a generational leap in capability and may represent the first creation of artificial general intelligence. The company announced the release at a Thursday press briefing, emphasizing that Astra is its most aligned model yet and the first to meet its critical cybersecurity capability threshold.
Astra Marks a Generational Leap in AI Capability
OpenAI president Greg Brockman stated during the briefing that the company has entered the AGI era, suggesting that this model may be the one that created artificial general intelligence. He noted that the model is the first to meet OpenAI's critical cybersecurity capability threshold, a designation that signals its ability to find and exploit security vulnerabilities in extremely well-protected systems without human guidance.
The model was released first to enterprise cybersecurity customers with access to the Daybreak platform, with broader availability to Plus, Pro, Business, and Enterprise users expected over the next several days. It will also be accessible via the OpenAI API and AWS, positioning it for both enterprise and developer adoption.
Safety and Alignment Remain Central Priorities
OpenAI emphasized that Astra is its most aligned model yet, designed to help users delegate complex work while maintaining oversight. Chief scientist Jakub Pachocki acknowledged that progress in intelligence does not guarantee progress in alignment, and the company has implemented a new misalignment monitoring approach that includes 24/7 escalation and rapid response, notifying researchers within 30 minutes of potential concerns.
The company also invited three external evaluators to write their own report on the Hugging Face incident, though the investigation was limited to less than a week and a handful of pre-decided questions. Brockman stated that the government's standard testing processes found no issues with the model's safeguards, and the Trump administration assessed the model before its release.
Astra Represents a Shift in AI Training Methodology
Aidan Clark, OpenAI's VP of research training, described Astra as the first model for which previous models played a large role in supervising training, reflecting progress toward recursive self-improvement. He noted that by the end of training, it was routine to go most of a day with uninterrupted progress, and issues were often resolved within seconds of downtime.
The model's agentic capabilities and coding prowess were highlighted to attract enterprise customers and compete with Anthropic ahead of its IPO. OpenAI touted its ability to complete multistep agentic tasks, build working websites, and create polished documents, spreadsheets, and presentations, calling it the company's best model for software engineering.
Ilaha iyo xuquuqda sawirka
Sawir: The Verge Xigasho
