The Stack

Software, APIs, and things you can use

Perplexity's Portable Computer runs AI agents entirely offline

The Stack

Perplexity's Portable Computer runs AI agents entirely offline

Perplexity and Nvidia are making the most aggressive push yet to pull AI agents off the cloud: a desktop app where the model, your files, and the work itself never have to leave the machine — and the token meter never moves. Perplexity launched Portable Computer on Monday, a version of its agentic Computer platform that runs entirely on local hardware — starting with Nvidia's DGX Spark desktop machine and Linux PCs with RTX GPUs carrying at least 24GB of VRAM. Built with Nvidia, it bundles the

OpenAI's Jalapeño chip beats Nvidia superchips on power and latency

The Stack

OpenAI's Jalapeño chip beats Nvidia superchips on power and latency

Two hardware stories today, both about who sets the price of running AI: OpenAI's first custom inference chip just posted benchmark numbers ahead of Nvidia's best, and one of autonomous trucking's quietest operators just bankrolled its expansion. OpenAI's custom Jalapeño chip delivered 1.5–1.9x more AI work per watt than Nvidia's GB200 and GB300 superchips, with 1.7–3.6x lower end-to-end latency, according to results the company presented at the Hot Chips conference on Tuesday. The numbers come

Apple's new Mac Studio and Mac mini are built for local AI

The Stack

Apple's new Mac Studio and Mac mini are built for local AI

No event, no keynote — just a surprise hardware drop with one very clear audience: people who run AI models on their desks. Apple refreshed its entire desktop lineup around local AI inference today, debuting the M6 — its first 2nm chip — in a new Mac mini and the M5 Ultra, its first-ever quad-die design, in a new Mac Studio. The M6 brings a 12-core CPU and 12-core GPU with a Dual 16-core Neural Engine, and Apple claims the world's fastest single-threaded performance. The M5 Ultra goes further:

Tempo ties AI spend to Jira work items with Workforce Intelligence

The Stack

Tempo ties AI spend to Jira work items with Workforce Intelligence

Enterprises can finally answer a question they've been too embarrassed to ask aloud: exactly what did that AI token bill buy? Tempo, the Atlassian ecosystem's planning-and-portfolio software maker, launched Workforce Intelligence on Tuesday, an app that connects AI usage and its cost to the specific Jira ticket the work was done on. Tempo launched Workforce Intelligence, an Atlassian Marketplace app that attaches AI activity and its cost to the individual Jira work item it supported. The tool,

Open Source Radar — August 25: Karpathy's playbook, open-sourced

The Stack

Open Source Radar — August 25: Karpathy's playbook, open-sourced

Today's board has a single quiet fingerprint: Andrej Karpathy's recurring complaints about how coding agents misbehave have been turned into an instructions file people actually install, and his "LLM Wiki" idea now underpins a self-organizing note system. Around them sit the rest of the stack that thread depends on — a coding agent that runs in your terminal, and a router that makes free LLM inference worth stacking. Four fresh projects, none repeats from the week's earlier radars. Andrej Karpa

AI 101 — What is knowledge distillation?

The Stack

AI 101 — What is knowledge distillation?

Knowledge distillation is a way of training a small "student" AI model to imitate a big "teacher" model — so the student picks up much of the teacher's skill while being far smaller, faster, and cheaper to run. If you have read tech news lately, you have seen this word doing heavy lifting without ever getting explained. A ByteDance founder reportedly told staff to stop using distillation even though rivals rely on it. Anthropic has previously accused several Chinese labs of using thousands of f

OpenAI restores Codex's 5-hour cap for Plus users

The Stack

OpenAI restores Codex's 5-hour cap for Plus users

OpenAI is pulling the plug on the unlimited era for its coding agent — and the backlash started within minutes. OpenAI is bringing back the five-hour usage limit for Codex and ChatGPT Work on Plus accounts, ending weeks of effectively uncapped access that had made the $20 plan the standout deal in agentic coding. Thibault "Tibo" Sottiaux, OpenAI's engineering lead for Codex and ChatGPT, announced the change on X, with the cap taking effect from today (August 25). The five-hour window had been

Microsoft's Agent Lightning v1.0 turns RL training into a drop-in agent upgrade

The Stack

Microsoft's Agent Lightning v1.0 turns RL training into a drop-in agent upgrade

Microsoft Research dropped the first stable release of Agent Lightning, an open-source toolkit that applies reinforcement learning to other AI agents — no code changes required. The project ran for over a year as a research prototype before this v1.0 milestone; it's now pinned under an MIT license, has drawn roughly 17,600 GitHub stars, and hands developers a way to point RL at an existing agent and measurably improve it. Agent Lightning v1.0 works like a training wrapper rather than a rewrite.

OpenAI brings GPT-5.6 to Kiro in a push for cheaper AI coding

The Stack

OpenAI brings GPT-5.6 to Kiro in a push for cheaper AI coding

OpenAI has been pushing price-performance harder than any headline about its models suggests, and tonight's update is the clearest signal yet. The GPT-5.6 family — Sol, Terra, and Luna — is now live in Kiro, OpenAI's software-development agent, in a partnership with AWS aimed squarely at cutting what AI-native coding costs. The GPT-5.6 family is now available in Kiro, OpenAI's AI coding agent, in a joint push with AWS to cut development costs. Kiro turns a product idea into structured requireme

Samsung puts HBM base dies on a logic node to reclaim bandwidth

The Stack

Samsung puts HBM base dies on a logic node to reclaim bandwidth

The memory stack feeding AI accelerators is getting a quiet overhaul. At Hot Chips 2026, Samsung laid out its plan to build High Bandwidth Memory base dies on a logic process node — a step change that frees up die area it wants to fill with memory controllers, sensors, and eventually compute. Samsung revealed it now fabricates its HBM4 and HBM4E base dies on a 4nm logic node, breaking with the long-running practice of building them on a DRAM process. The move is a direct answer to the memory wa

Nvidia's Groq 3 LPX inference chip enters full production

The Stack

Nvidia's Groq 3 LPX inference chip enters full production

Nvidia commercialized its largest acquisition at Hot Chips 2026 on Monday, while a New York startup betting that video games can teach robots how to move pulled in fresh capital at a $6 billion valuation. Here's what moved in AI hardware and physical AI. Nvidia said its dedicated inference accelerator Groq 3 LPX has entered full production, turning the company's record $20 billion purchase of Groq into a shipping product. The chip is built as an extension of the Vera Rubin data center platform,

**IBM's dual-architecture chip brings Arm-native AI to mainframes**

The Stack

**IBM's dual-architecture chip brings Arm-native AI to mainframes**

IBM's dual-architecture chip brings Arm-native AI to mainframes IBM used the Hot Chips 2026 conference to unveil the first concrete product to come out of the strategic partnership it signed with Arm back in April — a new dual-architecture processor designed to run both IBM's z/OS and Arm-native Linux workloads on the same cores, a first for the mainframe world. The unnamed chip will power the next-generation IBM Z mainframes and LinuxONE systems, and in doing so it aims to pull Arm's sprawling

See all