Latest AI News
Nurix AI Acquires Verloop.io, Expanding AI Agents Across Voice and Chat
Following the acquisition, Verloop.io’s team and operations will be integrated into Nurix AI.
View
Semiconductor Fabs to Benefit as Govt Scraps Import Duty on Critical Electronic Components
The waiver of basic customs duty covers essential inputs such as capital goods and equipment for producing lithium-ion cells, including machinery to enhance domestic production.
View
Mistral Bets on Physical AI With 8 Bn Parameter Autonomous Robot Navigation Model
French AI startup Mistral expands into robotics with Robostral Navigate, an 8B navigation model that enables autonomous movement using a single camera.
View
SpaceXAI Launches Grok 4.5, Narrows Gap with GPT 5.5 and Opus 4.8
SpaceXAI has priced Grok 4.5 at $2 per million input tokens and $6 per million output tokens.
View
This startup thinks robotics is about to have its ChatGPT moment
Loading the player… Before OpenAI’s GPT-3 ushered in the era of foundation models, companies built specialized natural language processing models from scratch, training each on large amounts of task-specific data. Today, most organizations start with a general-purpose model like OpenAI’s GPT series, Claude, or Llama and then fine-tune or prompt it to solve their specific needs. Pim de Witte, CEO ofGeneral Intuition, thinks embodied AI will follow a similar pattern. Rather than collecting huge real-world datasets to build specialized robot models, he argues the industry should focus on better quality datasets that can produce foundation models capable of transferring intuition about movement and interaction across many environments. “A lot of companies right now are doing lots of specialized work focused on individual embodiments, individual environments, and individual robots,” de Witte told TechCrunch on arecent episode of Equity. Much of that work will become redundant soon, he argues, with the emergence of general models like the one General Intuition has been developing and deploying. “The generalization of the model itself is the product,” he said. “The fact that it has a base level of reasoning about space and time is going to be the reason why people stop collecting hundreds of thousands or millions of hours of real-world data. Because the reality is, you only need a few minutes.” General Intuition built its own such foundation model after training on millions of hours of video game data, including information like what buttons on a controller a human pushed and when. Both de Witte and General Intuition’s lead investor, Vinod Khosla, argue the action data is the key to developing a human-like intuition for spatial-temporal reasoning. The startup last monthraised $320 million at a $2.3 billionvaluation on the back of that thesis. The company has demonstrated that its current model is capable of both playing a video game for hours and powering a quadrupedal robot — the latter after fine-tuning it on just eight minutes of real-world robotics data. “The fact that [the robot] was actually able to zero-shot on just the front camera, with no other sensors, in the office with dynamic objects being introduced and people walking by was a very big surprise to us,” de Witte says. “I think it’s a sign of what’s to come.” The end game for General Intuition isn’t to build robots itself, but to become the foundation model of physical AI, a base model for other robotics companies to build upon for their own machines. Or, as de Witte put it: “We’re not gonna build a self-driving car company. We’re gonna make it 10 times easier for the next person to build a self-driving car company.”
View
SpaceXAI releases Grok 4.5, which Elon describes as an ‘Opus-class model’
SpaceXAI has released its latest model, Grok 4.5 — the first since the company went public several weeks ago. Ina blog postpublished Wednesday, SpaceXAI characterized its new release as a workhorse that can tackle all of the typical tasks that the AI industry has sought to automate: coding and app-building, office and clerical work, research, writing, and other forms of routine knowledge work. Grok can supposedly do all this for less spend, too, as SpaceXAI says that its model has “twice greater token efficiency” than other leading models. If it carries through to real-world use cases, that efficiency would be a big advantage for SpaceXAI, since the cost of tokens has been a growing concern for AI consumers. The company released benchmark metrics Wednesday that appeared to show Grok’s competitiveness with other top models from SpaceXAI competitors, although just short of best-in-class: Ina poston his social media platform X (which is a subsidiary of SpaceXAI), founder Elon Musk compared the model to Opus, Anthropic’s LLM designed for intensive and complex tasks. “Based on strong positive feedback from customers in our beta test program, @SpaceXAI will make Grok 4.5 available to the public tomorrow. It is an Opus-class model, but faster, more token-efficient and lower cost,” wrote Musk in his post on X. Musk later added: “Our internal assessment is that Grok 4.5 is roughly comparable to Opus 4.7, but much faster. The combination of capability, faster speed and lower cost is what makes it competitive.” SpaceXAI says that its new model costs $2 per million input tokens and $6 per million output tokens. That’s quite competitive, if Grok’s capabilities match SpaceXAI’s rhetoric. Opus 4.7, by comparison, costs $5 per million input tokens and $25 per million output tokens. OpenAIhas tiered costsfor different model versions: Sol, its most expensive, costs $5 for 1 million input tokens and $30 for 1 million output tokens, while its least expensive, Luna, costs $1 for 1 million input and $6 for 1 million output tokens. It’s a big week for AI model releases. OpenAI isplanning to releaseGPT 5.6, its latest, most powerful model, on Thursday. The release of that model had previously been limited by the Trump administration, due to concerns about its security implications. OpenAIhas called itits “strongest model yet.”
View
Google’s deepfake detector system used to debunk McConnell hoax pic
Google’sSynthIDsystem has been used to debunk a high-profile AI-generated hoax image, in a rare but significant win for the system. Earlier this week, a picture circulated online that seemed to show Kentucky Senator Mitch McConnell covered in tubes in a hospital bed in a state of extreme distress. The image was shared widely onRedditandX, but by Wednesday, the revered fact-checking siteSnopeshad debunked the image, noting that, when checked, the image registers as containing the SynthID watermark designed by Google to identify AI-generated pictures. In short, the watermark worked exactly as it was supposed to in a win for anti-deepfake technology. Senator McConnell’s health has been the subject of intense speculation since he checked into the hospitalafter an emergency call on June 14. Since that time, he’s been largely absent from the public eye, fueling speculation that his health may be failing. In this case, however, the evidence proved to be entirely fake. Launchedat Google’s I/O developer conference in 2025, SynthID works as an invisible signature, visible to SynthID algorithms but designed to be unnoticeable to the casual observer. Because the signature is built into the image itself, it survives even when an image is screencaptured across multiple platforms, as the McConnell image was. SynthID’s main limitation is that it can only be used when an image-generation tool actively participates in the program. Gemini models have included the watermark since the program launched in 2025. OpenAI joined in May 2026, as part ofa broader effort to fight malicious image generation. Anthropic does not participate in the program. Users can check if images contain the watermark by asking a Gemini model or uploading them toOpenAI’s public image verification tool.
View
Lovable reportedly in talks to double its valuation to $13.2B
Lovable, a Swedish vibe-coding startup, is in talks to raise $300 million at a valuation of $13.2 billion — exactly double the$6.6 billionvaluation the company achieved last December,Sifted reported. Menlo Ventures, a firm that announced its latest$3 billion fundlast month, is expected to lead the round, according to the report. The less-than-three-year-old startup hit$500 millionin annualized revenue run rate in June. Lovable’s users include founders, individual designers, and salespeople building websites and e-commerce storefronts. The company also sells its vibe-coding tool to large enterprises, including Workday, Asana, and Nvidia. Vibe coding, which allows users to build software simply by describing it, is by far the most popular and lucrative use case for AI. Other high-profile vibe-coding startups include Replit, valued at$9 billionin March, and Factory, a startup that helps enterprises develop AI agents, whichraised $150 millionat a $1.5 billion valuation in April. Meanwhile, Cursor, which offers vibe coding for developers, was acquired by SpaceX for$60 billionlast month.
View
Your gaming data could be the secret to AGI, according to this Bezos-backed startup
When it comes to achieving artificial general intelligence (AGI), large language models just don’t have what it takes. Models like ChatGPT and Claude are great at text, but they’re less skilled at understanding how things actually move through space and time — an essential skill for producing intelligence that generalizes. That gap, it turns out, might be filled by gaming data. That’s the bet behind General Intuition, a Bezos-backed, New York-based startupvalued at $2.3 billion that just closed a $320 million roundwith Coatue, Eric Schmidt, and researchers at MIT and Google DeepMind joining its list of investors. On this episode of TechCrunch’sEquitypodcast, General Intuition CEO Pim de Witte joins Rebecca Bellan to dig into why world models trained on gaming data might be the next big leap in physical AI, how the company spun out of gaming platform Medal TV, and where the ethical red lines are when your models could end up being used for defense applications. Listen to the full episode to hear more about: Subscribe to Equity onYouTube,Apple Podcasts,Overcast,Spotifyand all the casts. You also can follow Equity onXandThreads, at @EquityPod.
View
These AI startups are growing revenue at faster and faster rates
As companies old and new rush to capitalize on AI, many AI startups say that their revenue is not just growing, but also rapidly accelerating, hitting their next milestones in shorter timeframes. The following list of startups have reported a pattern of such flywheel growth. One thing to note is that the underlying metrics used by these companies differ, even if they are using the term “ARR.” Some may be referring toannualized recurring revenue (ARR),or revenue under contract from a paying customer but not yet billed. Some are referring to annualized run-rate revenue, or projecting annual income by calculating 12 months of revenue that continues at the rate of the most recent month. Others are referring to “committed ARR,” or signed contracts from customers that are not onboarded yet. In the case of Gusto, it reported actualtrailing 12-month revenue. Nevertheless, each of these startups, listed in reverse chronological order to when their ARR growth was made public, reports that their revenue growth is accelerating, however they are defining it. To be sure, there are many more fast-growing AI startups than we’re naming here, but we are limiting this list to the companies hitting revenue milestones at ever-faster rates. Mercor:On Monday, Brendan Foody, co-founder and CEO of Mercor,announcedthat the company has crossed $2 billion in gross annualized revenue as of June — just four months after reaching the $1 billion milestone. The less-than-three-year-old firm, which hires domain experts to train and refine AI models,saidthat it reached a $500 million run rate in September. Anthropic:In recent months, this model maker’s revenue has been at such a historic velocity that it has mesmerized the entire AI sector. In late May, Anthropic announced that itcrossed $47 billionin revenue run rate, a milestone that came less than two months after the company reported that its revenue run rate surpassed$30 billion. The company said it reached a$9 billionrevenue run rate in late 2025, up from a reported$4 billionin July 2025. Sierra:After reaching its first$100 millionin ARR in seven quarters, Sierra — which builds customer service AI agents for enterprises — says it took just two more quarters to add another $100 million, co-founder and CEO Bret Taylorannouncedin late May. Glean:In May,Glean announcedthat it crossed $300 million in ARR. While it took the seven-year-old enterprise AI startupnine months to doubleits ARR from $100 million to $200 million, the company says it needed just six months to grow that metric from $200 million to $300 million. Gusto:The 14-year-old HR tech startup announced in May that its revenue accelerated in each of the last five quarters. The company, which was last valued at $9.3 billion in early 2022, also reported that it surpassed$1 billionin trailing 12-month revenue. Gusto’s revenue surge shows that it’s not only AI-native companies that are seeing their top-line growth supercharged by integrating the technology. Clio:This 18-year-old provider of legal practice management software saw its revenue take off sharply after embedding AI into itsoffering in 2023. The company surpassed $200 million in ARR in mid-2024, doubled that figure by late last year, and recently announced that its ARR reached$500 million.
View
Prime Intellect raises $130M Series A to help enterprises build their own AI agents
Prime Intellect, a startup that provides computing power and specialized software tools that help companies build AI agents, has raised a $130 million Series A at a $1 billion valuation. The massive round was led by Radical Ventures, with participation from Nvidia Ventures, Intel Capital, Dell Technologies Capital, Iconiq, and a long list of angel investors who are founders of notable companies, including Aravind Srinivas (Perplexity), Aaron Levie (Box), Winston Weinberg (Harvey), Jeff Wang (Cognition), and Brendan Foody (Mercor). Founded in 2024, Prime Intellect’s goal is to give organizations capabilities to train their own agentic systems without relying on frontier AI labs. While this mission would have been hard to achieve just a few years ago, the rise of reinforcement learning techniques, which iteratively reward successful task completion and penalizes errors, can allow companies to become their “own AI lab” by refining models for specific business tasks. Although it is now possible to bypass closed AI labs, the underlying infrastructure remains so complex that most companies lack the expertise to assemble these pieces into a production-ready system. That’s where Prime Intellect comes in. The startup has developed what it calls a “full stack” for AI agent development, which includes compute access, a reinforcement learning framework, and evaluation tools. Prime Intellect’s platform functions like a marketplace, providing modular access so customers can pick and choose the specific tools they need without being locked into an all-or-nothing system. “They’ve stitched this together and built it in such a way that they’re operating at the frontier in a way that’s affordable,” said David Katz, a partner at Radical Ventures. He added that while others offer bits and pieces, Prime Intellect is unique in providing the capabilities of a top-tier AI lab as a “one-stop shop” for development. The startup’s approach has attracted customers like Ramp, Zapier, and Flapping Airplanes, who pay the startup for a hosted version of its tools. This rapid adoption has propelled the company to an annualized revenue run rate of $100 million. This growth is driven by the tangible results. For example, Ramp used Prime Intellect to build an agent that helped the fintech find answers inside spreadsheets. “The result beat the frontier models on accuracy while running at faster speeds and a fraction of the cost,” Ramp’s co-founder and co-CEO Karim Atiyeh said in a statement. Another key factor driving Prime Intellect’s growth is the recent realization by companies that building on top of frontier labs carries a number of risks. Companies increasingly don’t want to provide their proprietary information to OpenAI and Anthropic due to the risk of losing control over their data. They are also wary of depending on models that can be suddenly turned off, as happened with Anthropic’s Fable last month. “How do I know that I’m not working with a company that is going to try to replace me and generalize to what I’m doing,” Katz said. “All of these things are causing people to think, ‘How do I own my own enterprise intelligence and not have these risks’.” Prime Intellect co-founder and CEO Vincent Weisser believes enterprises are looking to move away from closed source frontier models, and his company provides the infrastructure to make that transition possible. “It shouldn’t just be a few nerds in a glass tower in San Francisco that have the capability to train AI models,” he told TechCrunch. “It should be every enterprise, every nation state.”
View
OpenAI releases new voice models for more natural live conversations
OpenAI today released new conversational models, called GPT-Live-1 and GPT-Live-1 mini, claiming that they sound more natural and can handle turn-taking better. These are full-duplex models, meaning they can speak and listen at the same time, allowing users to interrupt naturally and enabling features like live translation. The company is also replacing its current Advanced Voice Mode in ChatGPT with GPT-Live-1 mini by default. Users of paid tiers will be able to access the larger GPT-Live-1 model. The previous model combined a speech-to-text model to transcribe speech, a large language model to generate responses, and a text-to-speech model to deliver the final answer. The company said in a press briefing that the new models solve issues like interrupting users while they’re talking and not having enough intelligence to answer questions. OpenAI’s new models will send the query to its latest text models like GPT-5.5 for search, reasoning, or agentic capabilities while continuing the conversation. OpenAI also showed that the model can stay silent for a long time and absorb the context of the conversation until it’s called upon. Plus, as the new voice mode has access to newer GPT models, it can also present some information in a visual format. Other startups like Monogram,which raised $40 million in seed funding from DST and Lux Capital, are also leaning into visual responses to make assistants more interactive. The company said the new voice mode in ChatGPT is designed to have longer conversations. During the briefing, ChatGPT Voice’s product lead, Atty Eleti, said he has had 30- to 40-minute-long conversations with the voice feature during walks. OpenAI thinks that voice could be the primary interface to computing for complex work. Reports have suggested that it could launcha pair of earbuds with AI capabilities this year. However, it didn’t provide any information on hardware products. “Over time, we think this will also unlock the ability to use voice as a kind of primary interface to computing, and to manage increasingly complex long-running agentic work. The kind of amazing use cases that we see people using Codex and ChatGPT to accomplish, we think voice can be the future interface to all kinds of work,” Eleti said. OpenAI has worked on bolsteringvoice-basedfeaturesover the past few years to make ChatGPT’s voice mode sound more natural. The company said that more than 150 million people talk to ChatGPT using features like Voice and Dictation. Rivals are also attempting to make assistants more expressive. BothAppleandAmazonhave updated their assistants to be more conversational with better context handling. Startups likeSesame, founded by Oculus co-founder Brendan Iribe and Ankit Kumar, also launched AI assistants with more natural conversation while completing tasks in the background. OpenAI is moving in the same direction, aiming to let users talk to its assistant hands-free for a longer time. Despite its claim that the new voice mode sounds more natural, the company emphasized that it’s not aiming to make this an AI companion. It noted that the new models have safeguards built in to give age-appropriate responses to teens and provide resources if the conversation turns to topics like self-harm. The new voice mode still needs work. During the demo, when the company showed its live translation feature in Hindi, the assistant had a heavy American accent and spoke in Hindi that was unnatural sounding and had slightly bookish tone. The company said the new mode is optimized for “most spoken languages” but didn’t specify which ones.
View