OpenAI has released the GPT-5.1 family of models. GPT-5.1 Instant is a high-volume model that uses adaptive reasoning - it can decide when to think longer on hard prompts. GPT-5.1 Thinking can vary its deliberation for each question, so simple tasks are faster and complex work gets deeper analysis. GPT-5.1 Auto routes queries between models. ChatGPT has gained stronger personalization controls - users can now pick between different personalities.
The Steam Frame is a new wireless VR headset with pancake optics, a 110-degree field of view, and inside-out tracking. It will not require a gaming PC to work. The Steam Machine is a more powerful but less portable follow-up to the Steam Deck designed to fit into a standard-sized shelving unit to connect to a display. The Steam Controller is designed to replace a mouse. All of the new products will be released in 2026. No prices have been announced yet.
Waymo has started offering robotaxi rides on freeways for customers in Los Angeles, Phoenix, and the San Francisco Bay Area. It is the first company to offer fully autonomous freeway rides without a human specialist in the car to fare-paying riders in the US. Freeways are more controlled environments in some respects, but the higher speed poses a much higher risk. Cars need to be able to predict what happens much farther down the road. Waymo has announced plans to more than double the number of cities it operates in.
Dark excitons are exotic light-matter states found in atomically thin semiconductors. They usually remain invisible because they emit light very weakly. Researchers have found a way to make these previously hidden light states shine brightly and controllably. This discovery opens up a new frontier for faster, smaller, and more energy-efficient technologies.
Homebrew 5.0.0 is now available. The most significant changes since 4.6.0 are download concurrency by default, official support for Linux ARM64/AArch64, timescales for deprecating macOS Intel, and the removal of macOS Gatekeeper bypass behaviours. This post lists major changes and deprecations since 4.6.0. Homebrew has loosened its criteria for accepting software.
This repository contains a collection of software development kits for integrating Stripe with large language models and agent frameworks. Agent Toolkit integrates Stripe APIs with popular agent frameworks through function calling. The AI SDK integrates Stripe's billing infrastructure with Vercel's ai and @ai-sdk libraries. Token Meter integrates Stripe's billing infrastructure with native SDKs from OpenAI, Anthropic, and Google Gemini without any framework dependencies.
Creativity is not like productivity - you can't just work harder and expect creativity to appear. Sprinkle in new inputs instead, and see what new outputs appear. This post lists 12 inputs that help increase creativity.
Coinbase plans to leave Delaware and reincorporate in Texas because Texas' legal framework is more predictable and efficient. Delaware has long been the legal home of most big companies in the US, but the state's courts appear to be biased against founders and their boards. Texas has capitalized on executives' frustrations with Delaware by establishing the Texas Business Court and passing laws to make it more difficult to sue board members at companies incorporated in Texas and more challenging for shareholders to file stockholder proposals against companies.
Template interpreters generate machine code for each bytecode op handler and create an instruction dispatch table that maps bytecode opcode to machine code to execute it, allowing developers to do interesting things like keep important/hot variables in dedicated registers or use the hardware call-stack instead of simulating the stack.
This week, Baseten released Kimi K2 Thinking on Model APIs. Kimi K2 Thinking rivals the leading closed-source agentic models and is engineered for complex reasoning and agentic workflows.With Baseten's API you can get an LLM that is smarter, faster and cheaper. The Kimi K2 Thinking API is built for production workloads with high uptime, scalability, and performance stability:
Anthropic is investing $50 billion to build AI data centers in Texas and New York with Fluidstack. This effort supports the US AI Action Plan and reflects a long-term push for domestic AI leadership.
Marble turns text prompts, photos, videos, or 3D layouts into persistent, editable, and downloadable 3D environments, available now with a free tier. This is the company's first release after raising $230 million and positions the company as the leading “world model” provider for gaming, VFX, VR, and robotics customers.
Surprisingly, the authors of "AI 2027" (who predict superintelligence within a decade) and "AI as Normal Technology" (who expect gradual diffusion like electricity) agree on many areas. They say most benchmarks will be saturated by 2027-2028 with AI matching expert humans, yet systems will still regularly fail at mundane tasks like "book me a flight to Paris" through 2029 due to long-tail errors. The world will look basically the same in 2029, even with AGI right around the corner.
OpenAI spent $5.02 billion on inference alone with Microsoft Azure in the first half of 2025. This pattern continued through the end of September. The company's inference costs have risen consistently over the last 18 months. OpenAI's inference costs easily eclipse its revenues.
The 2025 automation landscape is confusing. Many solutions can deliver quick wins, but also trap value in silos, create technical debt, and block visibility. This guide [10] provides a comprehensive comparison of leading automation technologies and platforms, highlighting what truly matters when orchestrating complex, cross-functional processes. Download the Camunda comparison guide.
AlphaProof reached silver medal performance in the International Mathematical Olympiad 2024. This post shares details about how the model works. AlphaProof demonstrates how test-time compute can be scaled to obtain improved answers over many orders of magnitude. The researchers developing it were entirely limited by how many TPUs they could get their hands on.
ERNIE-4.5-VL-28B-A3B-Thinking is a 28-billion-parameter multimodal AI model that only activates 3 billion parameters during operation. It has a sophisticated routing architecture that allows it to match or exceed the performance of much larger competing systems on several categories of tasks. The model's most distinctive feature is 'Thinking with Images', a capability that allows the AI to dynamically zoom in and out of images to examine details. The feature dramatically elevates the model's ability to handle long-tail visual knowledge. The model has been released under the permissive Apache 2.0 license, which allows unrestricted commercial use.
Microsoft aims to double its total data-center footprint over the next two years. Its newest AI super factory sits within a set of two-story structures in Atlanta. The site will contain hundreds of thousands of Nvidia GPUs and dedicated high-speed connections to other locations. Part of Microsoft's Fairwater network of AI centers, it spans more than 1 million square feet across 85 acres.
Software interfaces force their way of thinking on users. AI allows you to bypass all these tools - you just have to feed it the right data, and it can complete complex tasks. Doing this right can result in you getting rid of almost all of your productivity tools. This article shows readers how to get started, along with examples of how AI can completely change a workflow.
The platform connects brands with voice rights owners, ensuring permission and fair compensation, and includes the voices of Michael Caine, Mark Twain, and Maya Angelou.
Anthropic divided eight of its researchers into two teams, one with Claude access, one without, and tasked them with programming quadruped robots to fetch beach balls.
OpenAI is resisting legal efforts by the New York Times to obtain millions of private ChatGPT conversations, citing user privacy and data security risks.
SoFi is the first nationally chartered US bank to launch direct in-app crypto trading for Bitcoin, Ethereum, and Solana, leveraging its banking charter and 12.6 million users who already trust it with checking and loans. The company has operated Lightning Network remittances with Lightspark since August, targeting the $740 billion global remittance market, where users send dollars that convert to pesos using Bitcoin rails invisibly. SoFi plans to launch a proprietary stablecoin by mid-2026, positioning crypto trading as a complement to its remittance strategy rather than a standalone product. 60% of crypto-owning members prefer licensed bank custody over exchanges.
JPMorgan is rolling out JPM Coin on Coinbase's Ethereum L2, Base, introducing a new class of interest-bearing deposit tokens backed by bank funds. These tokens enable instant, 24/7 settlements and will be accepted as collateral on Coinbase, following a successful pilot with Mastercard and B2C2. Other major banks like BNY Mellon and HSBC are pursuing similar products. This move underscores how Ethereum has become the preferred settlement layer for global finance.
Aerodrome's critics misunderstand token economics by ignoring velocity. 49% of AERO supply is locked as veAERO, creating ~5.6% net circulating inflation comparable to gold's 2-3% annual rate despite higher gross emissions. The protocol generates $185M in annualized revenue with zero cash burn since emissions are equity distribution rather than expenses, mirroring Google's early stock-based compensation strategy that showed accounting losses while preserving cash. veAERO holders capture 100% of protocol revenue plus rebases that offset dilution, while Aerodrome commands 60% DEX market share on Base with $425M total revenue since launch and accumulation from Coinbase, Circle, and Animoca Brands.
Public ICOs have replaced airdrops as the dominant token launch method after airdrop farming turned distribution events into exit liquidity for sybil attackers. Projects are now raising hundreds of millions at pre-launch valuations that still offer retail profitable flips. PUMP raised $600M at $4B FDV while XPL presalers captured 6x returns despite drawdowns, and upcoming launches MegaETH and Monad are raising $50M at $900M FDV and $200M at $2.5B FDV, respectively, with MegaETH oversubscribed 27.8x. This ICO surge redirects capital from existing altcoin holders into new launches, creating a headwind for high-FDV vaporware projects and accelerating dispersion between coins with monetary premium or cashflow versus narrative-driven tokens.
Market positioning has reset following October's washout, but crypto remains the worst-performing cross-asset class despite selective altcoin strength in DePIN (+22%), L2s (+13%), and mid-caps (+15%), driven by a weekend bounce. Bitcoin sits at $105k, 16% below all-time highs, a range where historical data shows BTC outperforms alts 54% of the time and explains why recent dinocoin rallies in FIL, ICP, and FET quickly lost momentum. Supportive macro conditions, including rate cuts and ended QT, haven't translated to broad crypto flows. The next volatility drivers are expected to come from US regulatory and political headlines rather than positioning shifts, as majors must reclaim leadership before sustained altcoin breadth returns.
Ethereum's stablecoin market cap hit an all-time high of $167B, cementing its position as the leading stablecoin ecosystem. Meanwhile, Binance Smart Chain flipped Solana, pushing it to fourth place after a $2B decline in circulating stables. With over 80% of non-centralized stablecoin liquidity now on Ethereum L1 and L2 (Arbitrum and Base holding $8B combined), capital is increasingly consolidating around Ethereum's DeFi and payment infrastructure.
For users with a moderate-sized portfolio, one path to accumulating wealth is participating in upcoming presales and allocating capital to new platforms and projects to farm potential token launches and airdrops.
Google Photos is rolling out major updates powered by its new Nano Banana AI model, including "Help Me Edit" for prompt-based photo modifications and the ability to transform still images into videos or artistic styles. The "Ask Photos" feature is available in iOS in over 100 countries, offering 17 language options. Android users in the US and India have access to AI templates that provide ready-made editing prompts, such as professional headshots or fashion photoshoots. A new "Ask" button enables quick Gemini conversations about specific images. All features will become available to users this week.