Skip to main content

icnlive

Menu
Menu
icn.lıve
WATCH LIVE. THINK BUSINESS.
Menu
X-twitter Linkedin-in Instagram Youtube

© 2026 ICN.LIVE
  • OpenAI

OpenAI’s executive departures continue as its top data center chief exits

OpenAI’s executive departures continue, and the newest one hits a corner of the company that decides how fast its AI can grow.

Trending AI

Explore Section:

TECH
AI
Capital
Politics
Banking
Health
Dubai Is Ready With Free Autonomous Taxi Service
  • Apollo Go

A single line on the RTA’s social feed told residents their next ride might arrive with no one at the wheel. Dubai is ready with a free autonomous taxi service, and the offer began in two neighbourhoods that hug the coast. The Roads and Transport Authority opened public rides in Umm Suqeim and Jumeirah, both close to the beaches, on July 15, 2026.

Riders reach the vehicles through two apps rather than an RTA channel. Uber customers pick the Autonomous option and receive a WeRide car when one is free nearby. Apollo Go users book inside that company’s own app. Fleet allocation depends on how many vehicles sit inside the active zone at the time.

The split behind the wheel is worth noting. Tawasul Transport handles dispatch and operational control for WeRide taxis booked through Uber. Dubai Taxi Company manages local fleet needs for Apollo Go. The technology firms run the driving systems. The licensed operators handle the ground work under RTA supervision.

This Dubai driverless taxi rollout removes something the earlier trips carried. When Uber first offered WeRide rides in December 2025, a vehicle specialist sat inside. The current free service drops that specialist and completes the move to fully autonomous operation.

From testing to public roads

The path here was slow and deliberate. RTA started commercial operations on March 30, 2026, after trials on set roads. By September 2025, more than 60 vehicles from Apollo Go, WeRide and Pony.ai were already running in Jumeirah and Umm Suqeim.

Apollo Go Dubai began with 50 RT6 vehicles used for testing and data collection. Agreements signed in April 2025 made Dubai the company’s first operating market outside mainland China and Hong Kong. Each RT6 carries 40 sensors and detectors. By April 2025, Apollo Go had logged more than 150 million kilometres of safe driving and over 10 million autonomous trips across several cities.

WeRide ran about 150 autonomous vehicles across the Middle East by December 2025, including more than 100 robotaxis. Anyone asking how to book driverless taxi Dubai rides will find the answer sits inside apps they already use.

What the cars actually do

The autonomous taxi Dubai fleet leans on artificial intelligence, high-definition maps and deep learning. Onboard systems read the road in real time and make driving choices without a person. The software handles intersections, signals, pedestrians and nearby cars while following road rules. These vehicles share open roads with live traffic.

The bigger target

Every robotaxi Dubai adds points toward one number. Dubai’s Self-Driving Transport Strategy aims for autonomous operation across 25 per cent of all journeys by 2030. The strategy covers several transport modes, not taxis alone.

Mattar Al Tayer, Director General and Chairman of the Board of Executive Directors at RTA, named licensing, infrastructure and operating rules as core requirements back in April 2025. Dubai’s population passed four million residents by December 2025, which lifts demand for taxis and app-based mobility.

The free window has limits Dubai has not spelled out. RTA has not disclosed operating hours, wider route boundaries, or an end date for free pricing. Published plans place 100 vehicles in the first phase and up to 1,000 Apollo Go units within three years. Later reporting suggests a Dh5 fare has since appeared on Apollo Go, a sign the free period may be short. Coverage will widen as demand, service quality and regulatory readiness allow.

  • 3 min read
Mira Murati's Inkling AI model
  • Mira Murati

Mira Murati’s Inkling AI model arrived this week, and it shifts who gets to shape a frontier system. Thinking Machines Lab posted the full weights on Hugging Face. You can download them, run the model on your own hardware, and rebuild it for your work. That reach sets it apart from the closed flagships sold by OpenAI, Anthropic, and Google.

What Inkling is

Inkling is a Mixture-of-Experts model with 975 billion total parameters. It calls on about 41 billion for any single task, which keeps a very large system faster and cheaper to run. The model handles a context window of up to 1 million tokens. It reads text, images, and audio, and reasons across all three. Thinking Machines trained it on 45 trillion tokens of text, images, audio, and video.

The company is blunt about where the model lands. In its own words, Inkling is not the strongest model available today, open or closed. The pitch runs a different way. Murati’s team wants you to own a base you can adapt, not rent a locked system you cannot see inside.

Why Mira Murati’s Inkling AI model targets builders

Mira Murati’s Inkling AI model is built to be changed. Thinking Machines made it available for customization on the Tinker fine-tuning platform on launch day. You point the model at your own data, train it for your domain, and keep the result. The company also added an Inkling Playground in the Tinker console, so you can chat with the model before committing to a training run.

Controllable thinking effort is the other lever. You can dial how long the model reasons, trading speed for depth depending on the task. A lighter preview, Inkling-Small, runs 12 billion active parameters and aims for strong results at lower cost and latency.

Who Murati is and why it counts

Murati left OpenAI as chief technology officer in 2024. Months later, she founded Thinking Machines Lab, which raised a $2 billion seed round at a $12 billion valuation before it shipped a single product. Inkling is the lab’s first in-house model, built in under a year and trained on NVIDIA GB300 NVL72 systems.

Her open-weight AI model choice carries a history. At OpenAI in 2019, the lab held back the full GPT-2 over misuse fears. Murati now signals a case-by-case path: release openly when the risk looks manageable, hold back when it does not. Open weights this time do not promise open weights next time.

What it means for you

The stakes reach past the research world. If you run a business, an open-weight AI model you can fine-tune keeps your proprietary knowledge in-house instead of feeding it through someone else’s API. Thinking Machines points to work with hedge fund Bridgewater, where a fine-tuned open model scored 84.7 percent on a financial reasoning test. That figure came from the two companies’ own evaluation, not an independent one, so weigh it as a first-party claim until others confirm it.

For developers, Mira Murati’s Inkling AI model widens the menu. You get a large, multimodal base you can inspect, adapt, and deploy through providers like TogetherAI, Fireworks, and Baseten. The Mixture-of-Experts model design keeps running costs down while the parameter count stays high.

Inkling is the first in a planned family from Thinking Machines Lab. More models will follow. Whether each future model ships with open weights is not settled yet.

  • 3 min read
Muse Spark artificial intelligence model
  • Mark Zuckerberg

Muse Spark artificial intelligence model now carries a price tag, and developers must pay for access. Meta released the upgraded version on Thursday, opening a public preview to United States developers. The company charges $1.25 per million input tokens and $4.25 per million output tokens. Every new account receives twenty dollars in free credits before pay-as-you-go billing starts. You can read this move as Meta finally selling access instead of giving models away.

Alexandr Wang leads the effort, and he calls the pricing aggressive next to rival lab offerings. His team built the model to handle coding work and long chains of agent tasks. Meta Superintelligence Labs trained it on real-world software problems across large enterprise code bases. Wang told CNBC the update marks the best coding and agent performance Meta has shipped. Rivals now face a cheaper option built by a company with enormous computing capacity.

The Meta Model API sits at the center of this shift toward paid developer access. Developers sign up through a portal, test prompts, compare outputs, and prototype their own integrations. Meta limits access to its own properties for now, skipping third-party marketplaces like OpenRouter. Some early partners already hold API keys, and new users enter a waitlist for entry. Replit, Cline, and Box rank among the first companies building on the new system.

Muse Spark artificial intelligence model sets a new price floor

Muse Spark 1.1 pricing lands below Anthropic’s Claude Sonnet 4.6 on both input and output. The rate still runs above cheaper tiers such as GPT 5 mini and Claude Haiku 4.5. Zuckerberg framed the cost as one of the lowest available to developers right now. In my assessment, price alone will not decide which lab wins the coding market. Quality, reliability, and developer trust matter as much as the number on the invoice.

Meta claims strong benchmark results, including wins over Google’s Gemini 3.1 Pro in some areas. The AI coding model handles bug diagnosis, feature builds, and large-scale code migrations. It supports a context window of one million tokens for long-running technical sessions. Engineers can run it as a lead agent or as a subagent inside larger systems. Mark Zuckerberg said, “Muse Spark 1.1 is strongest at agentic performance, tool use, and computer use.”

Wall Street keeps pressing Mark Zuckerberg for returns on enormous artificial intelligence spending commitments. The company spends like its hyperscaler peers, yet it owns no cloud infrastructure business. Meta plans to launch one, and paid model access opens a second revenue line. Earlier Llama releases went to the open source community without any charge to users. Wang says an open source variant remains in development, though he gave no release date.

What Muse Spark artificial intelligence model means for your stack

Muse Spark artificial intelligence model gives you another vendor inside a crowded developer market. Meta trained the release to work with popular agent harnesses developers already run daily. Wang points to health research as one use case, from web searches to academic papers. Your team should test output quality against cost before moving any production workload over. Meta faces a hard climb, and the Muse Spark artificial intelligence model carries heavy expectations. Developers now decide whether the Muse Spark artificial intelligence model earns a permanent slot.

  • 3 min read
Load More Economy News