Connecting to the Living Graph…

Products & Launches

New AI product launches and feature releases from OpenAI, Anthropic, Google, and emerging startups. First coverage of ChatGPT updates, Claude features, Gemini releases, and tools changing how we work.

AI Research Funding & Business Products & Launches Big Tech Startups Open Source Policy & Ethics Opinion & Analysis

A stylized pelican pedaling a bicycle rendered in vibrant digital art, displayed on a laptop screen inside a cozy…

Products & Launches

Qwen 3.7 27B Draws Best Local-Model Pelican, Simon Willison Says

Qwen 3.7 27B, as a 17GB GGUF, drew Simon Willison's best local pelican-bicycle image. Signals on-device generation crossed a quality threshold.

x.com/19h ago/3 min read

image-generationhardwarelocal-ai

A developer inspects a wall of code on a large monitor, highlighting a vulnerability marker beside a performance…

Products & LaunchesBreakthrough

Zhipu GLM-5.3 Beats Anthropic Mythos on CyberGym, Lags ExploitBench

Zhipu's GLM-5.3 edged Anthropic's Mythos on CyberGym detection (84.5% vs 83.8%) but trailed badly on ExploitBench (54.4% vs 78%), exposing an exploitation gap.

scmp.com/1d ago/3 min read

benchmarksai modelscybersecurity

Data center racks with glowing server lights, a sleek AI interface dashboard showing a token speed meter at 750…

Products & LaunchesBreakthrough

OpenAI's Ultrafast Mode Hits 750 Tokens/s on GPT-5.6 Sol

OpenAI launched Ultrafast mode for GPT-5.6 Sol at 14x speed and 750 tokens/s, powered by Cerebras. Preview limited to select customers, targeting latency-sensitive enterprise workflows.

techcrunch.com/1d ago/3 min read/Multi-Source

ai inferencehardware accelerationopenai

Developer monitors live GitHub star count surging past 35,000 on a laptop screen displaying DeepSeek's open-source…

Products & Launches

DeepSeek Open-Sources Harness (dsh) With Plugin Architecture

DeepSeek open-sourced DeepSeek Harness, a plugin-based agent harness that crossed 35k GitHub stars in hours. It treats adapters, tools, and session logs as swappable plugins, addressing context-assembly pain points.

x.com/1d ago/3 min read/Multi-Source

open sourceai frameworksdeepseek

Close-up of a laptop screen displaying colorful code editor with a glowing AI assistant icon in the corner

Products & Launches

Gemini 3.7 Flash Ships Improved Long-Horizon Coding

Google released Gemini 3.7 Flash with improved long-horizon coding and PDF understanding. No benchmarks or pricing disclosed.

x.com/1d ago/3 min read/Multi-Source

geminiai modelsgoogle

Demis Hassabis in a suit speaking to officials in a White House meeting room, gesturing toward charts on AI safety

Products & Launches

Hassabis Pitched Frontier AI Safety Body to Trump Officials

Hassabis pitched an industry-funded AI safety standards body to Trump officials before leaving DeepMind. The proposal would test models for national-security risks via federal agencies and national labs.

x.com/2d ago/3 min read

ai safetyai policyregulation

Engineer in a data center pointing at a rack of TPU servers with glowing blue status lights, while a colleague…

Products & Launches

Google's 7-Year-Old TPUs Still Run at 100% Utilization, Says Vahdat

Google's 7-8 year old TPUs still run at 100% utilization per Amin Vahdat, citing Jevons Paradox. Efficiency gains drive demand, keeping all chips busy.

x.com/2d ago/3 min read

ai infrastructurehardware lifecycletpu

OpenAI Codex Hits 15M Users, Marking AI Coding Reset

Products & Launches

OpenAI Codex Hits 15M Users, Marking AI Coding Reset

OpenAI's Codex surpassed 15 million users, per @kimmonismus, signaling a reset in AI coding assistants. The milestone pressures rivals like GitHub Copilot.

x.com/2d ago/3 min read

ai codingcompetitionopenai

A laptop screen displays a graph visualization with connected nodes, while a person's hand points at the diagram…

Products & Launches

Neo4j-backed course teaches agentic knowledge graph construction

Neo4j and Andreas Kollegger released a free course on agentic knowledge graph construction using Google's ADK, promising better grounding than RAG. The course teaches parallel processing of CSV and text into a linked graph.

x.com/3d ago/3 min read

knowledge graphsagentic aieducation

Meta Drops Muse Glimmer 30B Under Apache 2.0, First OSI L…

Products & Launches

Meta Drops Muse Glimmer 30B Under Apache 2.0, First OSI License

Meta released Muse Glimmer 30B under Apache 2.0, its first OSI-licensed open-weight model, marking a strategic shift toward true open-source AI.

x.com/4d ago/3 min read/Multi-Source

open sourcemodel releasemeta

Vercel Sandbox Adds MicroVM, Free Egress Firewall After K…

Products & Launches

Vercel Sandbox Adds MicroVM, Free Egress Firewall After Kimi, OpenAI Breaches

Vercel Sandbox now uses microVM isolation and a free egress firewall after Kimi and OpenAI incidents showed container isolation failing for frontier agents.

x.com/4d ago/3 min read

ai infrastructuresandboxingsecurity

Developer at a Mac laptop coding an inference engine, terminal with H3 logs visible, open-source community workspace…

Products & Launches

Redis Creator antirez Writes MiniMax H3 Mac Inference Engine

Redis creator antirez wrote a MiniMax H3 Mac inference engine, announced on X. Open weights enabled the port, highlighting community-driven hardware support.

x.com/4d ago/3 min read

open sourceappleai hardware

OpenAI's GPT-5.6-Cyber Answers 95% of Blocked Security Queries

Products & LaunchesBreakthrough

OpenAI's GPT-5.6-Cyber Answers 95% of Blocked Security Queries

OpenAI launched GPT-5.6-Cyber, answering 95% of security queries other models block, up from 57.3%. Found two Chrome zero-days.

the-decoder.com/4d ago/3 min read/Multi-Source

ai modelscybersecurityopenai

Engineer at a workstation demonstrates a dashboard with a latency spike graph and code commit details on a large monitor

Products & Launches

Anthropic Engineer Shows Claude Managed Agents for Server-Side AI

Anthropic engineer demoed Claude Managed Agents, a server-side harness with 90% lower P95 latency and an SRE agent that traced a P99 spike to a commit.

x.com/5d ago/3 min read/Multi-Source

anthropicai agentsdeveloper tools

Sam Altman: AI Token Usage Growing Exponentially

Products & Launches

Sam Altman: AI Token Usage Growing Exponentially

Sam Altman claims AI token usage grows exponentially, referencing a 6.5-year baseline. The remark signals compounding inference demand that pressures infrastructure scaling.

x.com/5d ago/3 min read

ai infrastructureinferenceopenai

Engineers in cleanroom suits inspect advanced semiconductor packaging equipment with optical modules and test…

Products & Launches

TSMC Pushes CPO, CoPoS; AUO Eyes Test Partner Role

TSMC advances COUPE and CoPoS; AUO seeks testing role. Report via @dnystedt, with Ennostar, Tyntek, GCS as potential supply chain partners.

x.com/5d ago/3 min read

advanced packagingco-packaged opticssemiconductors

A crowded UK tribunal hearing room with a judge at the bench, stacks of case files overflowing on desks, and a…

Products & Launches

AI Claims Flood UK Employment Tribunals, Backlog Hits 64K

UK employment-tribunal backlog hit 64,000 cases, up from 45,000, as workers use AI to draft claims free. Judges say AI filings overwhelmed the system, exposing a cost-asymmetry crisis.

x.com/6d ago/3 min read

legal techemploymentuk policy

A magnifying glass over a glowing circuit board with a Nvidia chip, symbolizing investigation into remote access

Products & Launches

US Probes Chinese Firms' Remote Access to Nvidia GPUs

BIS probes Chinese firms' remote access to Nvidia chips, a potential export-control gap. Smuggling is covered; remote rental may not be.

x.com/6d ago/3 min read

export-controlschinanvidia

Philippines' $40B BPO Sector Faces AI Job Erosion

Products & Launches

Philippines' $40B BPO Sector Faces AI Job Erosion

AI is targeting the Philippines' $40B BPO industry employing 1.9M people. Voice and data roles face the highest exposure, threatening 7% of GDP.

x.com/Aug 7, 2026/3 min read

voice aibusiness process outsourcingphilippines

A sleek creative software interface showing a 3D motion control panel, video timeline, and reference thumbnails…

Products & Launches

Seedance 2.5 Drops on Magnific With 3D Motion Control

Seedance 2.5 launched on Magnific with exclusive 3D motion control, 50 references, and 30-second generations. The release marks a productization step for ByteDance's video model.

x.com/Aug 7, 2026/3 min read

ai videomodel launchcreative tools

A round, doughnut-shaped smart speaker with a fabric finish sits on a wooden table, held by one hand, no screen…

Products & Launches

OpenAI's First Device: Doughnut-Shaped Speaker, No Display

OpenAI's first device is a doughnut-shaped, display-less smart speaker, per Bloomberg. It's battery-powered with moving parts for interaction, targeting voice-first AI computing.

x.com/Aug 6, 2026/3 min read

hardwareconsumer aiopenai

Products & Launches

OpenAI Ships GPT-5.6 Sol as Unified Reasoning Model

OpenAI unified ChatGPT reasoning under GPT-5.6 Sol for Plus/Pro, with Luna for free tiers. Internal eval shows 68% fewer factual errors.

x.com/Aug 6, 2026/3 min read

model releasechatgptopenai

Meta Ships Muse Code Beta, Its First Coding Agent

Products & Launches

Meta Ships Muse Code Beta, Its First Coding Agent

Meta released Muse Code beta, its first coding agent built on Muse Spark 1.2. The release enters a market led by GitHub Copilot and Amazon CodeWhisperer.

x.com/Aug 6, 2026/3 min read/Multi-Source

coding agentsmetaai models

Researchers in a lab huddle around a large monitor displaying a bar chart of benchmark scores, with sticky notes and…

Products & Launches

Agent Memory Benchmarks Get a Shared 5,000-Question Test

A 20+ institution consortium launched a shared agent memory benchmark with 5,000 questions and one answer model, targeting the attribution problem in memory startup claims. First rankings due mid-August.

x.com/Aug 6, 2026/3 min read

open-sourcebenchmarksai evaluation

Laptop screen showing a document parser interface converting a PDF file to markdown text, with a speed metric overlay

Products & Launches

anydoc: Sub-5ms Local PDF Parsing for AI Agents

Nick Camara announced anydoc, a local parser claiming sub-5ms markdown conversion for 10+ formats. Claims 100x speedup, but no benchmarks or code released yet.

x.com/Aug 5, 2026/3 min read

document parsingai agentsdeveloper tools

NVIDIA Vera CPU chip on a dark circuit board, labeled with BlueField-4 branding, surrounded by glowing blue components

Products & Launches

NVIDIA Vera CPU Claims 3.67x Storage Speed vs x86

NVIDIA's Vera Arm CPU claims 3.67x faster storage processing than x86, targeting Intel and AMD via BlueField-4 STX.

x.com/Aug 5, 2026/3 min read

arm cpunvidiadata center

Developer workstation with code editor and terminal windows showing Claude Code agent configuration files

Products & Launches

Matt Pocock Open-Sources .agents Skill Directory for Claude Code

Matt Pocock open-sourced his .agents directory for Claude Code, including /grill-me and /tdd skills. The release targets agent misalignment in production engineering workflows.

x.com/Aug 4, 2026/3 min read/Multi-Source

agent-toolsopen-sourcedeveloper-workflow

A legal document with highlighted email text sits on a desk beside a laptop and smartphone, evoking a corporate dispute

Products & Launches

Apple-OpenAI Trade-Secret Fight Turns on Offboarding Emails

OpenAI published Apple emails showing staff sought help from ex-engineer Chang Liu after his departure, undercutting Apple's offboarding claims while leaving core theft allegations intact.

x.com/Aug 4, 2026/3 min read

trade secretslegalai companies

NVIDIA Releases Nemotron VoiceChat, First Open Full-Duplex Speech Model

Products & Launches

NVIDIA Releases Nemotron VoiceChat, First Open Full-Duplex Speech Model

NVIDIA released Nemotron VoiceChat, claiming the first open full-duplex speech model with tool calling and barge-in. The move targets real-time voice agents, challenging proprietary APIs.

x.com/Aug 3, 2026/3 min read/Widely Reported

open-sourcespeech-ainvidia

A sleek AI dashboard interface displaying marketing analytics and revenue metrics, with a robot hand adjusting data…

Products & Launches

Helena: Self-Improving AI Marketer Claims $10M in Sales

Enrich Labs claims Helena, a self-improving AI marketer, generated $10M in sales across 20,000 businesses. Claims lack verification or technical detail.

x.com/Aug 3, 2026/3 min read

startup claimsai marketingautonomous agents