Elon Musk’s artificial intelligence (AI) startup xAI is set to launch its first model, Grok, according to multiple posts from the Tesla CEO this week.
In a post on X on Friday, Musk announced that the xAI Grok AI assistant would be included with X Premium Plus. He also reiterated the point in an additional post on Saturday, adding that users can subscribe to the upgrade tier of the social media platform tier for $16 a month.
The software is the first product to be launched by the company since Musk started xAI in July.
The @xAI Grok AI assistant will be provided as part of 𝕏 Premium+, so I recommend signing up for that.
Just $16/month via web. https://t.co/wEEIZNjEkp
— Elon Musk (@elonmusk) November 4, 2023
Later on Saturday, both Musk and xAI posted official announcements of the AI model, along with a page for Grok appearing on the company’s website.
Announcing Grok!
Grok is an AI modeled after the Hitchhiker’s Guide to the Galaxy, so intended to answer almost anything and, far harder, even suggest what questions to ask!
Grok is designed to answer questions with a bit of wit and has a rebellious streak, so please don’t use…
— xAI (@xai) November 5, 2023
Grok will have real-time access to the X platform, which Musk and the company have said is a huge advantage when compared with other generative AI models.
On the origin of the name “Grok,” Musk called one take early Saturday morning “accurate” after another user used GPT-4 to ask about the meaning of “grok,” to which the OpenAI foundation model responded that the word originates from the 1961 novel “Stranger in a Strange Land” by Robert A. Heinlein.
The book’s use of the term, according to the AI model, refers to a Martian word “that means to understand something or someone so completely that the observer becomes a part of the observed—to merge, blend, intermarry, lose identity in group experience. It’s a deep, almost metaphysical understanding.”
The rivaling language model also detailed the modern use of the word “Grok,” which has found a place in the tech world to “suggest a profound level of comprehension of software programming, technology, or a complex subject matter.”
Accurate
— Elon Musk (@elonmusk) November 4, 2023
Musk also said that Grok would have more of a sense of humor and sarcasm than other generative language models, even sharing a few screenshots from discussions with the AI. You can see his screenshots below.
Credit: Elon Musk | X Credit: Elon Musk | X
Musk also noted in a response to Whole Mars Blog that the threshold for what Grok will tell you, “if pushed, is what is available on the Internet via reasonable browser search.”
The Premium Plus tier on X comes as a range of tiers set to be offered on the platform, including two lower tiers called Basic and Basic Plus. The company also says on its support page for the tiers that pricing on Premium will start at $3 per month or $32 per year, based on localized pricing.
Although X and xAI are not the same company, it’s not abnormal for Musk’s businesses to work together. Some examples include the Boring Company’s use of Tesla vehicles in its underground loops and Musk’s previous hopes for utilizing OpenAI at Tesla.
Update 11/5/23: Added official release information from xAI.
Elon Musk’s X is seeing fewer downloads, but usage is hitting all-time highs: study
What are your thoughts? Let me know at zach@teslarati.com, find me on X at @zacharyvisconti, or send your tips to us at tips@teslarati.com.
Tesla’s reason for Starlink integration on Cybercab might surprise you, as the company’s Head of AI, Ashok Elluswamy, finally shed some light on the reason they are putting a satellite internet terminal on its ride-hailing-geared vehicle.
On Monday, Tesla officially confirmed that it would integrate Starlink V5 terminals into Cybercab vehicles, something many Tesla fans had figured the company would do, as the vehicle is primarily geared toward giving rides without any passenger intervention.
Starlink V5 directly integrated in Cybercab pic.twitter.com/FxzTtzjB6I
— Tesla (@Tesla) July 20, 2026
The ability to access the internet would allow riders to work or play in the car with their devices. It seemed like a more-than-reasonable feature to add to the Cybercab, which made its way off the production lines for the first time earlier this year.
Tesla reveals first vehicle model to receive Starlink integration
However, the move is not for the rider, as Elluswamy confirmed on Monday night. Instead, it’s actually for Tesla to be able to have a constant connection to the cars in the Robotaxi fleet so it can troubleshoot issues, contact riders, or resolve other issues.
Elluswamy said:
“It is still not required for safe operation of the vehicle. Connectivity is primarily meant for navigation, customer service and, in general, fleet management.”
It is still not required for safe operation of the vehicle. Connectivity is primarily meant for navigation, customer service and, in general, fleet management.
— Ashok Elluswamy (@aelluswamy) July 20, 2026
Many initially assumed the option of constant connectivity would be enabled on the Cybercab for passenger entertainment or work. With the Cybercab, passengers won’t be doing anything but enjoying the ride, so it seemed more than logical that they would be hanging out with Starlink internet access as an amenity.
However, Tesla’s primary concern with Robotaxi is safety, and nailing these first unsupervised rides is a crucial step to setting a good narrative on how effective driverless transportation can be.
Being able to get in touch with passengers or a vehicle if something is wrong is a crucial part of the overall experience, and preventative measures are being taken by Tesla to ensure a smooth process, even in the worst-case.
Tesla has expanded its Robotaxi program in Florida to include two new cities: Tampa and Orlando.
This marks the second and third cities to be added to the company’s available locations for autonomous ride-hailing in the Sunshine State, joining Miami, which was the first Florida city to offer Robotaxi rides.
Tesla announced the addition of Orlando and Tampa to the Robotaxi program on Tuesday morning. The cities now join Austin, Dallas, Houston, Miami, and the San Francisco Bay Area as locations where Tesla can operate its Robotaxi platform:
Robotaxi now in Tampa & Orlando! pic.twitter.com/fYtmXgJq3O
— Tesla Robotaxi (@robotaxi) July 21, 2026
These rides are unsupervised, as AI Head Ashok Elluswamy confirmed the suite in Florida is operating without safety drivers or anyone within the cabin to assist with operation.
Orlando Tesla Robotaxi Operation
The geofence in Orlando covers a prominent irregular shaded zone on the map, roughly 4-6 miles across in key dimensions, so it likely measures somewhere between 25 and 45 square miles, which is comparable to other early Tesla launches in other cities.
It encompasses central and southern areas bounded by major highways including SR-417 and SR-528, including parts of the Orlando metro core, tourism-adjacent zones, and residential/commercial districts. This represents an initial targeted rollout in a tourist-heavy region, positioned for quick expansion via Tesla’s software updates.
Tampa Tesla Robotaxi Operation
In Tampa, the shape of the geofence is a shaded polygon covering key neighborhoods, explicitly including West Tampa, Tampa Heights, Hyde Park, and downtown Tampa proper, with boundaries along major roads and the Hillsborough River area.
This focuses on high-demand central zones and will offer tourists and citygoers rides without drivers.
Robotaxi Progress
Tesla has been operating Robotaxi since last June, when it launched in Austin. The geofences in most regions have already expanded several times since their launch last year, but the bigger complaint is vehicle availability. Tesla has been working to add more Robotaxi-enabled vehicles to its fleet.
The company still plans to utilize its Cybercab, a new vehicle that is being produced at Gigafactory Texas, for the Robotaxi suite alongside the Model Y, which has been the vehicle of choice for Tesla with early operations.
News
Tesla’s AI Chief just hinted at something big for FSD v14 lite owners
Tesla’s AI chief suggests the newest FSD v14 Lite build may finally go wide release.
Tesla’s head of AI, Ashok Elluswamy, noted on Sunday that the newest FSD v14 Lite build rolling out to Hardware 3 cars is likely the version that goes to wide release, the strongest signal yet that Tesla is near to closing out an early access phase that Hardware 3 owners have waited more than a year for.
Elluswamy made the comment in response to an extensive review from Tesla owner Zack, known on X as @BLKMDL3, who tested software version 2026.20.6.10 and detailed the changes in a lengthy post. “FSD v14 Lite (for Tesla AI3 hardware vehicles) review.
The update restarts a rollout that had stalled after its initial release. Tesla began pushing FSD v14 Lite to Hardware 3 early access drivers on June 29, bringing driving behavior learned on the newer Hardware 4 computer down to the more limited chip that has powered Tesla vehicles built between 2019 and early 2023. That release, as we covered in detail, gave roughly 4 million HW3 vehicles their first meaningful update since being frozen on version 12.6 in early 2025.
Tesla Full Self-Driving v14 ‘Lite’ Release Notes: new capabilities and features
The latest build adds features that bring Hardware 3 closer in line with what Hardware 4 owners already have. FSD can now start directly from park without a brake pedal confirmation, a change Zack called a small but meaningful quality of life improvement. The interface also picks up the blue “P” park icon, approaching destination alerts, and a dedicated Self-Driving app with streak tracking, all details previously exclusive to the AI4 branch of v14, as outlined in Tesla’s original release notes.
The stakes around Hardware 3 go beyond software polish. Tesla sold the Full Self-Driving package for years on the promise that every vehicle equipped with it had the hardware needed to eventually drive itself without supervision. That promise broke down during Tesla’s Q1 2026 earnings call, when Musk acknowledged HW3 cars could not run unsupervised FSD, prompting Tesla to offer trade-in discounts and hardware retrofits alongside the Lite software track.
Tesla confirmed HW3 can’t do Unsupervised FSD but there’s more to the story
Tesla has continued to frame v14 Lite as the primary path forward for the HW3 fleet, telling owners in April that international markets would follow the U.S. rollout once regulatory approvals came through. For now, HW3 owners in the early access group are the only ones running the new build. A broader rollout would mark the second major software delivery to the legacy fleet since Tesla first released FSD v14 to Hardware 4 vehicles, and the first sign since June that the Lite program is still moving rather than stuck in early access limbo.
Tesla’s reason for Starlink integration on Cybercab might surprise you
Tesla Robotaxi program expands in Florida to two new cities
