OpenAI Pushes AI Agent Capabilities With New Developer API
An anonymous reader quotes a report from Ars Technica: On Tuesday, OpenAI unveiled a new "Responses API" designed to help software developers create AI agents that can perform tasks independently using the company's AI models. The Responses API will eventually replace the current Assistants API, which OpenAI plans to retire in the first half of 2026. With the new offering, users can develop custom AI agents that scan company files with a file search utility that rapidly checks company databases (with OpenAI promising not to train its models on these files) and navigate websites -- similar to functions available through OpenAI's Operator agent, whose underlying Computer-Using Agent (CUA) model developers can also access to enable automation of tasks like data entry and other operations.
However, OpenAI acknowledges that its CUA model is not yet reliable for automating tasks on operating systems and can make unintended mistakes. The company describes the new API as an early iteration that it will continue to improve over time. Developers using the Responses API can access the same models that power ChatGPT Search: GPT-4o search and GPT-4o mini search. These models can browse the web to answer questions and cite sources in their responses. That's notable because OpenAI says the added web search ability dramatically improves the factual accuracy of its AI models. On OpenAI's SimpleQA benchmark, which aims to measure confabulation rate, GPT-4o search scored 90 percent, while GPT-4o mini search achieved 88 percent -- both substantially outperforming the larger GPT-4.5 model without search, which scored 63 percent.
Despite these improvements, the technology still has significant limitations. Aside from issues with CUA properly navigating websites, the improved search capability doesn't completely solve the problem of AI confabulations, with GPT-4o search still making factual mistakes 10 percent of the time. Alongside the Responses API, OpenAI released the open source Agents SDK, providing developers free tools to integrate models with internal systems, implement safeguards, and monitor agent activities. This toolkit follows OpenAI's earlier release of Swarm, a framework for orchestrating multiple agents.
Read more of this story at Slashdot.
Girl, 13, charged with murder after mother-of-three found dead 'with stab injury' following house fire
The girl, 13, who cannot be named for legal reasons, is also charged with arson with attempt to endanger life. She will appear at Northampton magistrates' court on Wednesday.
Teenage girl, 15, 'grabbed by two men who tried to drag her into a car while she was walking along street'
A 15-year-old girl was allegedly grabbed by two men who tried to pull her into a vehicle as she walked along Colebrook Street, Plymouth at around 6.55pm on Sunday.
Teen tells of moment man and his dog were crushed and killed 'when car crashed into her taxi and flipped'
19-year-old Madison Burrows, who witnessed a car reportedly flip onto a pedestrian and his dog, reveals says she will 'never get the image out' of her head.
Entire families including women and children have been found slaughtered as violence erupts in Syria, UN says
Entire families were wiped out during last week's massacre across Syria in which more than 1,200 people died, the UN human rights office has said.
Pictured: Grandfather, 57, who died in motorbike crash - as his wife of 40 years pays tribute to her 'best friend'
Brian Rudkin, 57, died when his Piaggio PX motorcycle collided with a black Nissan Terrano in the morning on March 2 on the A606 in the village of Langam, East Midlands.
Primark's £16 'pretty' pink duvet set that shoppers say they 'need'
It is perfect for pairing with pink cushions too
Body found in search for missing university student, 21, who vanished while walking home from the pub
Sebastian Sailes, 21, was last seen leaving the Black Horse Pub in Chippenham at about 11.30pm on February 27 and was captured on CCTV heading towards Lovers Walk.
Choose your own Patch Tuesday adventure: Start with six zero-day fixes, or six critical flaws
Microsoft tackles 50-plus security blunders, Adobe splats 3D bugs, and Apple deals with a doozy
Patch Tuesday Microsoft’s Patch Tuesday bundle has appeared, with a dirty dozen flaws competing for your urgent attention – six of them rated critical and another six already being exploited by criminals.…
British tourist, 38, faces up to 10 years in jail after being arrested with 38 kilos of cannabis in his luggage at Thai airport as he tried to board a flight to London
Daniel James Hallett, 38, from Hull , was detained by customs officials as he allegedly attempted to board the flight to London from Krabi International Airport in the south of the country on March 9.
Is NASA's science budget heading for a black hole?
Dare mighty things ... as long as we can afford it
COMMENT NASA could be in line for severe cuts to its science budget, with a 50 percent reduction floated by folk in the space industry. The consequences would, according to observers, be nothing less than catastrophic.…
Eight days later, Microsoft Outlook users still struggle on iOS devices
Cloudy email rises like a zombie, though its digital grave still marked by big red cross
Outlook.com users on iOS trying to access their messages via Apple Mail are still struggling more than a week after users first reported service disruption, and Microsoft still hasn't confirmed the root cause.…
CHRISTOPHER STEVENS reviews The Au Pair on Channel 5: The useless villain in this bonkers drama wouldn't have fooled Poirot
Sir David Suchet's deductive power are slipping. In his Poirot days, he'd never have failed to spot a body in bovver boots, laid out in front of him.
It's so unfair that being an ethnic minority might mean you're treated more leniently by a court. 'Two-Tier Keir' must act: PROFESSOR ANDREW TETTENBORN
Shabana Mahmood has known about the proposed changes since they first landed on her desk after her appointment last July. She sidestepped the issue then: it has now come back to bite her
Why Aussie F1 star Jack Doohan is already in danger of losing his job after just ONE race for his team - as he cops abuse from fans before Melbourne Grand Prix
As the F1 circus rolls into Melbourne for the season-opening Australian Grand Prix, rookie Jack Doohan is already in danger of losing his job with Alpine racing team.
Oil tanker jet fuel spill 'will cause significant toxicity to marine life', professor says - as cargo ship captain is arrested over inferno
US oil tanker Stena Immaculate was carrying 220,000 barrels of jet fuel in 16 segregated cargo tanks when it was engulfed by a huge fireball after the Portuguese ship, Solong, ploughed into it.
OnlyFans model who goes by the name 'The Irish Viking' appears on list of Irish tax defaulters
Matthew Gilbert, who has hundreds of thousands of followers on social media, was found by Revenue to have under-declared his income by €61,734 (£52,068).
Ex-Meghan Markle defender Katherine Ryan savages Duchess in brutal takedown as she accuses her of LYING about Prince Harry while 'cosying up to celebrities and acting humble'
Comedian Katherine Ryan has publicly laid into Meghan Markle, who she previously defended, in the wake of her much-derided Netflix series, With Love, Meghan.
WhatsApp 'banter' could get you the sack, tribunal rules
Workers who engage in 'banter' group chats with colleagues are not able to use privacy as an argument to avoid being disciplined by bosses.
ANSWERS TO CORRESPONDENTS: Was the word Nosferatu invented for the film?
Stoker was clearly influenced by Gerard's writings, and her description of Nosferatu is familiar to those who have read Dracula.