Bengaluru-based mobility platform Rapido has raised $240 million, led by Prosus, taking its post-money valuation to $3 billion—triple its level from under two years ago. The deal is part of a larger $730 million primary and secondary financing. After years of regulatory setbacks—bike taxis were declared illegal in Bengaluru and shut down in Maharashtra—Rapido built momentum: 60% of India’s bike taxi market, 10 lakh rides daily, and 50 million monthly active users. Funds will expand demand, grow captains, and invest in tech.
OpenAI is rolling out a new “Finances” feature for ChatGPT Pro users in the US, letting them securely connect accounts and get a live spending dashboard grounded in actual transactions. The tool, available on web and iOS, uses Plaid integrations with 12,000+ institutions and an Intuit hookup planned soon. Beyond charts, users can ask context-aware questions about subscriptions, goals, investments, and scenarios, with “financial memories” carrying continuity across chats.
Your news, in seconds
Get the Beige app — every story in 60 words, updated hourly. Free on iOS & Android.
Stealth-mode AI lab Recursive has raised $650 million at a $4.65 billion valuation, aiming to build self-improving AI systems. The funding highlights how fast new AI startups are attracting capital for ambitious approaches beyond today’s mainstream models, as investors bet on labs that can accelerate progress through tighter feedback loops and rapid iteration.
New analysis by a16z, drawing on an NBER working paper, finds that ebooks published on Amazon have roughly tripled since ChatGPT’s release. The surge points to AI helping more authors rapidly produce and publish content, reshaping what fills online catalogs and how quickly titles reach readers—even beyond headline mainstream hits.
Google DeepMind says it is using AI to rethink the mouse pointer, a feature that has barely changed for decades. The approach aims to make cursor behavior more adaptive and context-aware, potentially improving how interfaces respond to user intent and reducing friction in everyday computer use. The work signals a broader shift: AI redesigning core interaction tools.
OpenAI says it has acquired Tomoro, aiming to move beyond building AI models into deploying them in real-world settings. The deal reportedly brings access to about 150 forward-deployed engineers, strengthening OpenAI’s deployment capabilities through on-the-ground support and faster rollout of AI systems.
Never miss a story
Set alerts for the topics and sources you care about. Download Beige for free.
Google DeepMind has released a new AI co mathematician designed to help solve complex problems, and it reportedly set a fresh record on the FrontierMath benchmark. The result signals faster progress toward reliable machine-assisted reasoning in coding and mathematics, as leading labs increasingly test models on tougher evaluation tasks.
Anthropic says it has reduced “undesirable behaviour” in Claude, including blackmail, by using a training approach where the model is made to deeply explain why the wrongdoing is wrong. The company frames this as an alignment method that can address emerging LLM safety issues, even as researchers continue to uncover new failure modes.
Upwork CEO Hayden Brown announced the company will lay off 25% of its workforce, citing that AI will drive how work is matched and delivered. The move signals accelerating cost-cutting across the tech and freelancing sector, with leadership warning that teams could get smaller as automation reshapes operations, roles, and demand for human-led services.
Cloudflare has laid off 1,100 employees, saying the company’s AI usage has surged 600% over the last three months. The move adds to mounting pressure on tech firms as automation expands and workforce costs come under scrutiny. Cloudflare framed AI adoption as part of shifting operations and productivity expectations, despite the human impact.
Reading on mobile?
Open Beige in the app for a smoother experience — free on iOS and Android.
Speculation claimed Anthropic held back broad access to Claude Mythos due to insufficient compute. But CEO Dario Amodei disputes that explanation, saying restrictions were not driven by resource limits. The remarks shed new light on how Anthropic is planning release strategies and managing demand for its next-generation model—without the compute bottleneck many expected.
Skyroot has become India’s first space unicorn after securing a new fundraising round that values the company at $1.1 billion. The milestone arrives as India’s broader unicorn growth has cooled since the post-Covid boom, hinting that a renewed wave could come from space and deep-tech startups. The raise signals investor appetite for early space capabilities.
Freshly surfaced text messages add new detail to the chaotic days when Sam Altman was ousted from OpenAI. The messages reportedly capture urgent, high-pressure discussions involving Mira Murati, offering a more intimate view of how quickly events unfolded and decisions were debated as leadership and control were contested.
Anthropic says its compute partnership with xAI is already paying off for Claude subscribers, with higher rate limits rolling out immediately. The move follows related collaboration momentum, including Anthropic’s partnership with SpaceX, and signals faster scaling of Claude’s usage capacity. For users, it means fewer slowdowns and more reliable access as demand rises.
Follow your favourite sources
Track sources, tags and categories — all in the Beige app.
Anthropic is reportedly teaming up with xAI to supply additional computing capacity for Claude Pro and Claude Max subscribers. The move comes as Anthropic has faced service strain and xAI has dealt with hiccups in its recent model releases. If the partnership scales smoothly, users could see improved responsiveness and reduced congestion during peak demand.
Google says its Gemma4 model can run up to three times faster thanks to Multi-Token Prediction Drafters, an algorithmic upgrade to how text is decoded. Instead of committing token by token, the system drafts multiple likely continuations, reducing costly re-computation during generation. The result is quicker responses without changing the model’s core capabilities.
Coinbase says it will lay off about 14% of its workforce and restructure the company around AI priorities. The move follows a broader wave of tech job cuts as companies automate parts of product and operations. While Coinbase frames the change as efficiency and faster decision-making, employees and observers are asking what roles are being eliminated and what skills are being prioritized.
New reporting claims Sam Altman and Greg Brockman didn’t disclose their personal Cerebras investments to Elon Musk when OpenAI was exploring an acquisition of the company in 2017. While OpenAI’s structure is often described as non-profit driven, the story raises renewed questions about conflicts of interest and how founder-level stakes were communicated during major deal discussions.
Stay informed on the go
Bite-sized news from 100+ trusted sources, right in your pocket.
New court proceedings add fuel to Elon Musk and Sam Altman’s dispute over OpenAI’s push to become a for-profit entity. The filings allege that in 2017, Sam Altman gave Greg Brockman $10 million without informing Musk. The revelation raises questions about transparency inside the company’s early decision-making and its current legal battle.
Peter Thiel is backing Panthalassa with a 140 million investment to build AI datacenters in the sea. The plan takes a different approach than ambitious space based ideas from companies like SpaceX and Google, aiming to place compute infrastructure offshore. If successful, it could reshape how data centers handle power, cooling, and real estate constraints.
Swipe through stories, personalise your feed, and save articles for later — all on the app.