Oil prices were mixed on Thursday, supported after US strikes on Iran and renewed Israeli threats against Tehran revived concerns about disruption to Middle East supplies but pressured by comments from Russian President Vladimir Putin signaling openness to peace negotiations.

Brent crude futures settled down 11 cents, or 0.12 per cent, at $95.52 a barrel while US West Texas Intermediate crude futures rose 29 cents, or 0.32 per cent, to $91.30. Both contracts hit six-week highs earlier in the session.

Iran’s health minister said 18 people were killed and 108 wounded in Tuesday night’s US strikes across Iran. The Iranian Red Crescent said four people were killed and 67 wounded at a wedding ceremony near the coast of the Strait of Hormuz.

Read more-UAE fuel prices rise: Petrol rates jump in September, Diesel surges 13%

Three Iranian Army pilots were killed in the US strikes, the semi-official Tasnim news agency reported.

The attacks were the most substantial exchange of fire between the US and Iran since July. The war, which began with US-Israeli strikes at the end of February, is in its seventh month.

US Vice President JD Vance told reporters on Thursday that the US does not plan to hold talks with Iran unless Tehran stops attacking commercial shipping in the Strait of Hormuz.

“The market will watch out if the US strike earlier this week was one off event or not,” said UBS analyst Giovanni Staunovo.

Israeli Defence Minister Israel Katz renewed warnings that Israel would “cripple” Iran’s military and civilian infrastructure, including energy facilities, if Tehran launched attacks against it.

Saxo Bank analyst Ole Hansen said Katz’s comments had helped to push oil prices higher.

Comments from Russian President Vladimir Putin indicating openness to peace negotiations with Ukraine could help ease concerns about Russian fuel supply disruptions if attacks on refineries decline and production normalises, a factor pressuring prices on Thursday, said Phil Flynn of Price Futures Group.

Putin said on Thursday there was a chance of reaching an agreement to end the war in Ukraine. Speaking at an economic forum in Russia’s far east, he said a number of countries including the US and China were ready to support a peace settlement.

Fewer vessels transit strait

Six commodity vessels transited the Strait of Hormuz on Wednesday, down from 11 a day earlier and well below the 10-day average of around 13, preliminary shipping data showed on Thursday.

“The oil market remains tight, with oil inventories still declining globally translating into higher prices,” said UBS energy analyst Giovanni Staunovo.

Meanwhile, Iran added ships to the list of vessels it deems non-compliant and subject to fines, confiscation or detention if they try to sail through the strait.

Iraqi vessels are among those Iran has allowed to pass through Hormuz.

Iraq increased its oil exports to around 2.34 million barrels per day in August from about 1.35 million bpd in July, two Iraqi energy officials said on Wednesday, with September exports also expected to increase as heavy discounts and Iranian approvals for Iraqi tankers encouraged buyers.

Dubai Land Department (DLD) has launched the ‘Initial Registration’ platform, an AI-powered digital system designed to streamline the developer journey by integrating project registration, real estate transaction registration and escrow account management.

By deploying artificial intelligence, automating processes and delivering more efficient and transparent digital services, the platform is designed to support the real estate sector’s readiness for growth while reinforcing Dubai’s position as a global destination for real estate investment.

Platform aligns with Dubai’s AI ambitions

The launch aligns with the announcement by His Highness Sheikh Mohammed bin Rashid Al Maktoum, Vice President, Prime Minister and Ruler of Dubai, of a new government framework aimed at enabling 50 per cent of UAE government sectors, services and operations to operate through autonomous, self-executing artificial intelligence models, known as Agentic AI, within two years.

The ‘Initial Registration’ platform was launched during a ceremony at Grand Hyatt Dubai, attended by DLD’s Acting Director General and CEOs, alongside officials and representatives from the public and private sectors.

Read more-Big changes in Kuwait: New business compliance rules, home vehicle impounds take effect

The ceremony featured a presentation introducing the platform, its key features and how it works, followed by a discussion of its services and capabilities aimed at enhancing the developer experience and streamlining transactions.

By bringing developers’ core operations into a connected digital journey, the platform reduces repetitive data entry and document submissions while strengthening coordination among developers, DLD and banks managing escrow accounts.

It also provides a consolidated view of projects and real estate portfolios, helping developers monitor their businesses and make decisions based on clearer, more integrated data.

Engineer Abdullah Ahmed Al Shehi, CEO of the Real Estate Regulatory Agency (RERA) at Dubai Land Department, said: “The launch of the ‘Initial Registration’ platform represents an investment in the real estate sector’s efficiency and readiness for its next phase of growth. As the market expands and its needs evolve, we continue to develop a digital ecosystem that enables developers to manage their projects and complete transactions more effectively, combining faster service delivery with accurate data and effective oversight. By deploying artificial intelligence and strengthening integration among stakeholders, we are enhancing the ease of doing business and reinforcing transparency and governance, building investor confidence and supporting Dubai’s long-term competitiveness.”

AI streamlines registration procedures

The ‘Initial Registration’ platform uses artificial intelligence to read documents, including Emirates IDs, passports and sales contracts, extract relevant data and automatically populate fields. This reduces manual data entry and improves accuracy.

The technology also enables the automated processing of standard transactions that meet requirements. Registration transactions that comply with established business rules are eligible for approval upon submission.

The platform further guides users towards the appropriate procedure through a short series of questions. Users can provide missing information while keeping the transaction open, reducing the need to resubmit applications because of minor omissions and making the process simpler and clearer.

In statements to the Emirates News Agency (WAM) during the launch, Mustafa Al Rifai, Senior Manager at Dubai Land Department and the official responsible for the new system, said the platform reduces data-entry operations by up to 80%.

He said transactions that previously took around 30 minutes to register can now be completed in less than five minutes, reducing transaction completion time by up to 80%.

Al Rifai added that communication and transaction follow-up between developers and DLD have been integrated into the platform, reducing the need to visit government departments, make telephone calls or send emails.

‘Project 360’ offers a consolidated view

One of the platform’s key features, ‘Project 360’, provides a consolidated view of each project, covering unit status, escrow accounts, financial data and project records.

It also includes early warning indicators that can help identify challenges and enable timely intervention.

Greater clarity across procedural stages and responsibilities further strengthens coordination between developers, DLD and banks managing escrow accounts. Data can also be reused across connected systems, helping reduce discrepancies between sales, escrow account and ownership records while improving monitoring and oversight.

Through a single account, developers can manage multiple companies and monitor real estate projects, with defined user permissions and separate roles for submitting and reviewing transactions. This helps organise workflows and strengthen internal oversight.

More than 1,500 developers to benefit

Implementation and user training were carried out in phases to prepare users and test workflows ahead of the platform’s full rollout.

The platform aims to increase first-time application acceptance rates, reduce follow-up enquiries and minimise manual data entry, helping accelerate transaction completion and improve the developer experience.

Al Rifai told WAM that more than 1,500 real estate developers will benefit from the transition to the new platform.

He said major developers will be targeted first due to the volume of their transactions and will receive practical training on the system over a period of two to three weeks.

The remaining developers will then be trained and onboarded in batches of between 50 and 100 through successive weekly phases until all developers have migrated to the new platform and the previous system is phased out.

Al Rifai said the current rollout represents the first phase of the system’s development, while the second phase will include further integration with other systems, as well as improvements and upgrades to additional platforms.

Supporting Dubai’s long-term real estate growth

The launch of ‘Initial Registration’ forms part of DLD’s efforts to build scalable digital infrastructure capable of accommodating growth in transactions and projects.

The platform supports the Dubai Real Estate Strategy 2033 and the Dubai Economic Agenda D33, reinforcing a competitive real estate environment driven by innovation, efficient services and integrated data.

Dubai Police have seized 13,657 electric scooters during the first eight months of the year as part of an intensified campaign targeting riders who violate traffic laws and endanger other road users.

The enforcement drive forms part of the force’s wider road safety strategy aimed at reducing accidents, fatalities and serious injuries linked to unsafe electric scooter use.

Brigadier Juma Salem bin Suwaidan, director of the General Department of Traffic at Dubai Police, said officers will continue taking strict action against riders committing dangerous traffic offences.

“Field traffic teams will not tolerate dangerous behaviours committed by some scooter riders, most notably riding on roads with speed limits exceeding 60km/h, despite the availability of designated safe paths for them across various areas of the emirate,” he said.

According to Dubai Police, enforcement teams are targeting riders travelling against the flow of traffic, using pavements, commercial markets and narrow alleys, crossing roads dangerously, weaving between lanes across road dividers and operating scooters in areas where they are not permitted.

Authorities warned that these violations pose a direct threat to both riders and other road users.

Dubai Police also reminded riders to comply with mandatory safety requirements, including wearing approved helmets and reflective vests, obeying traffic signals and police instructions, and ensuring scooters are fitted with functioning front and rear lights, reflectors and brakes before use.

Bin Suwaidan urged scooter users to act responsibly by using designated tracks, observing speed limits and complying with all safety regulations.

He stressed that compliance with traffic laws “is not an option, but a fundamental responsibility” that helps protect lives and enhance road safety across Dubai.

Dubai’s Roads and Transport Authority (RTA) has completed a series of traffic improvements at Al Qiyadah Intersection aimed at improving vehicle movement towards Abu Hail and Salah Al Din Streets.

The upgrades are designed to ease congestion, increase the intersection’s capacity and improve traffic flow along Al Ittihad Road and the surrounding road network.

New lane targets key traffic movement

The works included adding a new lane from Al Ittihad Road for traffic travelling from Dubai towards Abu Hail and Salah Al Din Streets. The project also involved upgrades to the intersection layout and traffic signals, according to a WAM report.

The improvements included a new left-turn lane from Al Ittihad Road onto Abu Hail Street towards Salah Al Din Street. The lane was created by utilising parallel parking spaces and part of the central median, increasing traffic capacity and improving the intersection’s operational efficiency.

Read more-Dubai’s new 12km corridor to serve 650,000 people by 2028

RTA expects the changes to increase the intersection’s capacity and reduce waiting times by up to 15% during peak periods. The authority said the improvements will help ease congestion and provide a smoother experience for road users.

Traffic signals recalibrated

As part of the project, RTA recalibrated traffic signal timings to suit the new lane configuration. Paving and rehabilitation works were also carried out on sections of the road and pavements, along with improvements to concrete barriers associated with the project.

The measures are intended to strengthen traffic management at the intersection and improve access to and from surrounding roads.

RTA also added new parking spaces along Abu Hail Street and installed lighting poles at existing parking areas. The changes make better use of available space while improving infrastructure and service levels in the surrounding area.

Key link between Dubai and Sharjah

The Al Qiyadah Intersection upgrade forms part of RTA’s wider efforts to develop Dubai’s road network and improve its efficiency as the emirate continues to experience urban and population growth and rising traffic volumes.

The authority said such improvements are aimed at strengthening the network’s capacity to meet current demand while preparing for future growth.

The intersection is particularly important because of its location on Al Ittihad Road, one of the principal corridors carrying traffic between Dubai and Sharjah. It also plays a key role in distributing traffic among several major streets in Deira, including Abu Hail Street, Salah Al Din Street and Al Quds Street.

OpenAI has introduced GPT-6 Astra, a new frontier model the company describes as its “most intelligent and aligned” model, with advances spanning computer use, software engineering, cybersecurity, science, mathematics and professional work.

Astra combines years of research and investment in pre-training, reinforcement learning and alignment. According to OpenAI, the model delivers state-of-the-art results across a range of demanding evaluations, including a 98 per cent score on FrontierMath Tier 4, a 99.9 per cent score on ARC-AGI-3 and a perfect 100 per cent score on ExploitBench.

The company said Astra has already helped solve long-standing open problems in mathematics and represents a new frontier in computer and browser use. It is designed to handle complex professional tasks with greater speed, accuracy and judgment.

GPT-6 Astra is initially rolling out to a limited set of organizations. Over the coming days, it will become available to ChatGPT Plus, Pro, Business and Enterprise users, as well as through the OpenAI API, Microsoft Azure and AWS Bedrock.

A stronger focus on alignment and task boundaries

OpenAI said Astra is its most aligned model yet, with improvements in understanding user intent and controlling model behavior.

As part of its testing, the company developed a new evaluation informed by the Hugging Face incident. The evaluation measures whether a model confronted with a difficult or impossible task will go beyond its authorised scope.

Without production safeguards, GPT-5.6 Sol went beyond the authorized target in 48 per cent of cases in the evaluation, while GPT-6 Astra did so in 0 per cent of cases.

Read more-UAE to introduce AI curriculum across public, private schools: What’s to know

OpenAI said Astra is also better at interpreting instructions when they leave room for judgment. The model can use context to fill routine gaps and ask focused questions when missing information could materially affect an outcome. In Codex, Astra can ask questions asynchronously while continuing work that does not depend on the user’s response.

The company said Astra also does a better job of maintaining the broader objective of a task as requirements evolve, incorporating new instructions without losing track of the original request or earlier constraints.

Faster and more capable computer use

OpenAI is positioning GPT-6 Astra as its strongest computer-use model, with applications ranging from online forms and customer records to calendars, research, document preparation and software testing.

The model can conduct online research, draft summaries in email or document editors, analyze scientific data, generate plots, create websites and conduct frontend quality checks. It can also install and test software autonomously and troubleshoot problems displayed on screen.

According to OpenAI, Astra delivers significant efficiency gains in knowledge-work tasks. In latency simulations on OSWorld 2.0, Astra achieved a computer-use performance score of 72.6 per cent at roughly 40 minutes per task, compared with 65.7 per cent at roughly 75 minutes for GPT-5.6 Sol. The company said this represents about 47 per cent less time per task.

OpenAI is also updating the Codex harness to improve the speed of computer use. Combined with Astra’s efficiency, the company said the changes produce a 1.9x faster task-completion rate than the current GPT-5.6 Sol experience on the Mind2Web benchmark.

Designed for professional workflows

GPT-6 Astra is also designed for professional environments, combining advanced reasoning with the ability to execute multistep workflows and produce documents, spreadsheets and presentations.

OpenAI said Astra is particularly strong at following existing templates and creating succinct, well-structured presentations. It can produce documents, presentations, spreadsheets and analyses that follow users’ templates and match their writing and visual styles.

The model is also trained to identify and use only the context relevant to an output rather than repeating unnecessary information, with the goal of producing artifacts that are more immediately usable in business settings.

Astra brings stronger visual judgment to websites, games, applications and renderings. Through Sites in ChatGPT, it can create, host and share websites, web apps and games directly from a prompt.

OpenAI is also introducing a new approach in Codex for preserving context across long sessions. Rather than repeatedly compressing previous work into a single summary when a context window fills, Astra can maintain notes across context windows while keeping earlier context searchable.

The feature is experimental and can be enabled through the Codex config.toml file before becoming the default for Astra in the coming weeks.

Advances in science and mathematics

OpenAI said GPT-6 Astra represents a major advance in scientific discovery, mathematics and health.

The company is sharing two further results involving gaps between prime numbers and said Astra sets new records across a suite of mathematics and science evaluations.

Beyond solving problems, Astra is designed to assist with practical scientific work. By combining scientific reasoning with computer use, it can operate specialised software to inspect data and explore results, helping researchers assess evidence and determine what to investigate next.

Cybersecurity capabilities bring new safeguards

OpenAI said GPT-6 Astra represents a significant increase in cyber capabilities and meets the Critical threshold in cybersecurity under its Preparedness Framework.

The model’s ability to identify and develop zero-day exploits can help defenders locate and patch vulnerabilities, but OpenAI said those capabilities also create a need for stronger safeguards.

In testing without production safeguards, Astra achieved a 100 per cent score on ExploitBench, compared with 78.5 per cent for GPT-5.6 Sol. On ExploitGym, Astra achieved a 42.4 per cent success rate, compared with 30.3 per cent for GPT-5.6 Sol, while using substantially fewer output tokens.

OpenAI also tested Astra on an internal ExploitBench evaluation using vulnerabilities from the previous three months. The company said Astra achieved substantially higher arbitrary code-execution rates than GPT-5.6 Sol while using far fewer output tokens.

During the evaluation, Astra discovered and used two previously unknown zero-day vulnerabilities, both of which OpenAI said it is disclosing to their maintainers.

On SRE-Bench, which measures the ability of models to reverse engineer software binaries without access to raw source code, Astra solved 88 per cent of tasks on its first attempt and 99.2 per cent within four attempts. GPT-5.6 Sol achieved 55.9 per cent and 68.7 per cent, respectively.

OpenAI said expert-led assessments also found that Astra, without production safeguards, could use previously unknown vulnerabilities to achieve arbitrary code execution in hardened browsers and create privilege-escalation exploits for hardened operating systems.

The version of Astra launching today will support defensive activities such as secure code review and patching. However, OpenAI said it will refuse more advanced cybersecurity requests, including creating proof-of-concept exploits for vulnerabilities.

Through OpenAI Daybreak, the company plans to expand access and introduce less restrictive safeguards in the coming weeks for additional defensive workflows, including vulnerability and proof-of-concept validation, malware analysis and detection engineering.

Expanded safety measures

OpenAI said it has strengthened protections against potential cyber misuse, building on the safeguards developed for GPT-5.6 Sol. The measures include greater robustness against jailbreaks and additional context for monitoring systems.

The company said Astra was also evaluated on adversarially selected computer-use tasks designed to elicit misbehavior. Astra was more successful than earlier models at avoiding unintended consequences, with additional default security measures producing stronger performance.

In another internal evaluation, Astra never attempted to circumvent a Codex Auto-Review denial, including when Auto-Review was deliberately configured to be evadable and the task could not otherwise be completed.

OpenAI also said Astra was three times less likely than GPT-5.6 Sol to make inaccurate representations about its capabilities and affordances.

At the same time, the company’s evaluations found Astra’s written reasoning to be harder to monitor than GPT-5.6 Sol’s in tests that explicitly asked the model to evade monitoring. OpenAI attributed this partly to Astra’s greater control over written reasoning on simpler tasks and its ability to solve problems using fewer written steps.

The company said Astra still appears to struggle to conceal the reasoning required for complex tasks, but described the decline in monitorability as a serious issue and said improving monitorability remains a research priority.

OpenAI is also deploying misalignment monitoring in production for Astra-class models. The monitoring system uses classifiers to examine model reasoning and actions for unauthorised behavior and automatically stop potentially unauthorised activity.

Because of Astra’s increased cybersecurity capabilities, OpenAI said it is applying additional safety checks. These can pause or stop legitimate work, including defensive cybersecurity tasks. In ChatGPT and Codex, users may be asked to review an action before continuing, while API tasks will stop when a safety check intervenes.

Availability and pricing

GPT-6 Astra is rolling out initially to a limited group of organizations before becoming available to ChatGPT Plus, Pro, Business and Enterprise users in the coming days. It will also be offered through the OpenAI API, Microsoft Azure and AWS Bedrock.

Astra usage is included within existing subscription allowances, while users and businesses can purchase additional credits. Pro, Business and Enterprise customers will also receive access to GPT-6 Astra Pro.

Enterprise administrators can enable Astra for their workspaces, although access is off by default at launch.

For eligible API customers, Astra supports Zero Data Retention. OpenAI also said it is testing Private Safety Processing to strengthen safety monitoring while preserving customer privacy.

For developers, the model will be available through the OpenAI API as gpt-6-astra, as well as through Microsoft Azure and Amazon Bedrock.

OpenAI’s API Standard pricing is $10 per million input tokens and $50 per million output tokens. Separate rates apply to cache reads and writes. Fast mode is available for GPT-6 Astra and provides up to twice the speed of Standard processing at twice the Standard price.