metir
metir
Docs
Download on App StoreGet it on Google PlayLog inSign up

metir Blog

Stories, insights, and perspectives on artificial intelligence, productivity, and the future of work.

PyTorch
Google TPU
AWS Trainium
Google TorchTPU and Amazon TorchNeuron: Why Both Clouds Are Targeting PyTorch to Loosen Nvidia's Software Hold
Amazon released TorchNeuron on October 19 and Google open-sourced TorchTPU on October 20, 2026, both letting PyTorch apps run on custom chips without a rewrite. An analysis of what changes, what does not, and why the software layer is the real contest.
Oct 23, 20268 min read
DeepSeek
AI Funding
China AI
DeepSeek's Reported $15 Billion Round: How a 50 Billion Yuan Raise Became 100 Billion
Bloomberg reports DeepSeek raised about 100 billion yuan (roughly $15 billion), two days after the US ordered federal agencies to remove its technology. An analysis of how the round grew, who is reported to be in it, and how it compares with other Chinese AI labs.
Oct 23, 20268 min read
1Password
AI Agents
Credentials
1Password for Agents: How AI Agents Can Log In Without Ever Seeing Your Password
1Password launched 1Password for Agents on October 22, 2026, letting AI agents sign in to personal accounts after user approval without seeing credentials. An analysis of the approval flow, the delegation models it competes with, and the open questions for banks.
Oct 23, 20268 min read
Meta
AI Capex
Stock Structure
Meta's New Non-Voting Stock Class: How It Could Help Fund a $135 to $145 Billion AI Buildout
Meta confirmed on its Q3 2026 earnings call that it will create a new class of non-voting capital stock, distributed as a stock dividend, for capital raising, employee pay and acquisitions. An analysis of the structure, what is undisclosed, and how it fits Meta's AI spending.
Oct 23, 20268 min read
Microsoft
Healthcare AI
Diagnosis
Microsoft's MAI-DxO Diagnoses 83% of Hard Cases in a Science Paper: What the Numbers Do and Do Not Show
Microsoft AI published a Science study in which its MAI-DxO multi-agent system diagnosed complex NEJM cases at 83%, versus 36% for physicians working alone. An analysis of the design, the model-versus-orchestrator split, the cost figures and the limits of a simulation.
Oct 23, 20268 min read
OpenAI
AI Safety
Agent Monitoring
METR and Redwood Red-Team OpenAI's Internal Agent Monitor: Six Vulnerability Classes and What They Show
Redwood Research and METR published results from a three-week red-team of OpenAI's internal agent monitoring system. An analysis of the six vulnerability classes, the 75 to 94 percent coverage estimate, the compaction gap and what it means for agent oversight.
Oct 23, 20268 min read
Anthropic
Claude
Formal Verification
Claude Formally Verifies 123 Open-Source Libraries With 180 Million Lines of Rocq Proofs
Anthropic's Frontier Red Team and Theorem used Claude to prove security properties of compiled code across 123 open-source libraries. An analysis of the method, the hundreds of bugs found, the token costs and the limits of what formal verification guarantees.
Oct 23, 20268 min read
OpenAI
Funding
Valuation
OpenAI Reportedly Nears $30B Round at $1.25T Pre-Money, Below the $1.4T It First Pitched
OpenAI is reportedly working on a raise of up to $30 billion at a $1.25 trillion pre-money valuation, led by Coatue with UAE investors. What changed since the $1.4 trillion report, how it compares with Anthropic, and what is still unconfirmed.
Oct 23, 20267 min read
Meta
AI Capex
Data Centers
Meta's Filing Shows About $825 Billion in Future Commitments, Up From Almost $700 Billion in July
A Meta filing lists about $825 billion in future leases and contracts, mostly tied to AI compute and data centers. A breakdown of the $355.7 billion in leases and $454.5 billion in non-cancelable contracts, how it differs from capex, and what it means.
Oct 23, 20267 min read
Agentic AI Foundation
Agent Skills
MCP
Agentic AI Foundation Adds Agent Skills, Open Responses, OpenEnv and OpenSandbox: What the Standards Stack Now Covers
At AGNTCon on October 22, 2026 the Linux Foundation's Agentic AI Foundation took on four projects from Anthropic, OpenAI, Hugging Face and Alibaba, growing to ten. An analysis of what each standardizes and which layers of the agent stack remain open.
Oct 23, 20267 min read
Walmart
Agentic Commerce
MCP
Walmart Opens an MCP Agent Gateway With Spending Limits While Amazon Blocks Shopping Agents
Walmart launched an MCP-based Agent Gateway and Agent Profiles that let approved AI agents search inventory and buy for shoppers, with spending caps and approvals. A look at the design, the terms for agents, and the opposite stance Amazon has taken.
Oct 23, 20267 min read
Warp Sciences
Kevin Weil
Drug Discovery
Warp Sciences Raises $215 Million: Why a Former OpenAI Science Lead Is Betting on Robot Chemistry
Former OpenAI science lead Kevin Weil launched Warp Sciences with $215 million at an $865 million valuation, led by Lightspeed, to automate wet-lab synthesis with robots. An analysis of the make-and-test bottleneck, the reshoring angle, and what remains unproven.
Oct 23, 20267 min read
Meta
Earnings
AI Capex
Meta Q3 2026: $64.6 Billion Revenue, $135-145 Billion Capex Guide and a $25 Billion New Mexico Penalty
Meta reported Q3 2026 revenue of $64.6 billion (up 26%) while operating income fell 37% on a $9.89 billion legal charge, narrowed 2026 capex to $135 to $145 billion, and was hit with a $25 billion New Mexico penalty the same week. An analysis of what the numbers say about AI spending.
Oct 22, 20269 min read
Flapping Airplanes
Funding
AI Research
Flapping Airplanes Raises $1 Billion at a $6 Billion Valuation: What a Year-Old Lab With No Product Is Betting On
Flapping Airplanes, a roughly year-old AI research lab with no commercial product, raised $1 billion at a reported $6 billion valuation. An analysis of the round, the megakernel GPU framework behind it, and the data-efficiency thesis investors are funding.
Oct 22, 20268 min read
Anthropic
Claude Duet
AI Agents
Claude Duet: Anthropic's Proactive Opus 5.5 Agent, What It Does and What Remains Unproven
Anthropic opened a research preview of Claude Duet, a standalone proactive agent on Opus 5.5 with morning briefs, a cloud computer and 6,000+ connectors. An analysis of how it works, the approval model and how it compares with other persistent agents.
Oct 22, 20268 min read
Cybersecurity
PHP
Wiz
PHP2Shell: Wiz Used Gemini 4 Argon and GPT-5.6 Sol to Find Critical PHP Flaws Back to Version 5.4
Wiz Research disclosed two CVEs covering four PHP bugs, with severity up to 9.2 and impact on releases back to PHP 5.4, found with AI models. A timeline analysis of the 63 days from discovery to patch and what AI-assisted vulnerability research changes for defenders.
Oct 22, 20268 min read
OpenAI
Google Cloud
Model Garden
OpenAI Models Arrive on Google Cloud: What the Model Garden Deal Changes for Enterprise AI
Google Cloud adds GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna to Model Garden in private preview. An analysis of the billing, data-governance and multi-cloud implications.
Oct 22, 20268 min read
Anthropic
AI Policy
AI Safety
Anthropic's Draft Federal AI Safety Bill: How Court-Ordered Model Suspension Would Work
Anthropic has reportedly pitched a bill letting Commerce seek a court order to suspend a model posing imminent catastrophic risk. A neutral look at the mechanism and open questions.
Oct 22, 20268 min read
Mistral
NVIDIA
Physics AI
Mistral and NVIDIA Target Industrial Engineering: Physics AI, CAE Benchmarks and What They Test
Mistral and NVIDIA plan open datasets and agentic benchmarks for CAE and chip design. An analysis of the closed-loop workflow, the metrics, and why evaluation is the hard part.
Oct 22, 20267 min read
Anthropic
Claude
AI for Science
Claude Designed 17 Scorpion Toxin Binders in 64 Hours: What the Result Shows and What It Does Not
Anthropic's Claude ran a 64-hour protein design campaign that produced 17 lab-confirmed binders against scorpion neurotoxins, including two with single-digit nanomolar affinity. An analysis of the hit rates, the funnel, and the gap between binding and an antivenom.
Oct 21, 20269 min read
SpaceX
xAI
Earnings
SpaceX Q3 2026: The AI Segment Is Now Its Largest Business at $7.56B, and Capex Hit $19.6B
SpaceX's second public earnings report shows AI segment revenue of $7.56B, nearly triple Q2, a swing to $4.4B net income, and $19.6B of quarterly capex. An analysis of what drove the beat, what the analyst capex and debt projections imply, and the caveats.
Oct 21, 20269 min read
OpenAI
GPT-6.1
COBOL
GPT-6.1 Sol Scores 97.9% on OpenAI's COBOL Migration Test: What Legacy Modernisation Still Needs
OpenAI says GPT-6.1 Sol fully migrated 97.9% of programs in a COBOL-to-modern-language test, up from 5.4% for GPT-5.6 Sol, and GPT-6 Astra averaged 87.4% on MultiLangBench. An analysis of the benchmarks, their limits, and what banks and governments still have to verify.
Oct 21, 20269 min read
Nvidia
China
AI Chips
Chinese Tech Giants Order Over 3 Million Nvidia RTX PRO 5500 Cards: The $35B Demand and Supply Gap
Reuters reports ByteDance, Alibaba and Tencent ordered more than 3 million Nvidia RTX PRO 5500 cards worth over $35 billion, while Nvidia warns it cannot fill them. An analysis of the demand-supply gap, the inference angle, and the 2027 Rubin capacity trade-off.
Oct 21, 20268 min read
AI Agents
Agent Identity
Web Bot Auth
Only 2.3% of Legitimate AI Agent Sessions Carry a Verified ID, Arkose Labs Reports
Arkose Labs' Q3 2026 trust report finds 97.7% of legitimate AI agent sessions are unsigned, while agents reached 77.8% of customers' password-reset flows. An analysis of the data, the methodology caveats, and what Web Bot Auth adoption would change.
Oct 21, 20268 min read
Tenstorrent
AI Chips
RISC-V
Tenstorrent's $1.5 Billion Series E at a $6.2 Billion Valuation: What a RISC-V AI Chip Maker's Raise Signals
Tenstorrent raised $1.5 billion in a Series E at a $6.2 billion post-money valuation on October 19, 2026. An analysis of the round, the RISC-V Galaxy servers it will fund, and what it says about the AI hardware market beyond Nvidia.
Oct 21, 20268 min read
Previous
1...456...31
Next

Showing 101-125 of 769 posts

Experience AI That Works for You

Join thousands of users leveraging metir to boost their productivity with the world's best AI models.

Get Started Free
metir

metir is where work gets done. Hire your team of Orbs, AI teammates with their own computer, alongside your mail, meetings and workspace.

© 2026 metir. All rights reserved.

Product

  • Features
  • Pricing
  • Research
  • Docs
  • Blog
  • Enterprise

Company

  • Docs
  • Support
  • Careers

Legal

  • Terms of service
  • Privacy policy

Personalisation is powerful. Privacy is non-negotiable.

Status: All systems operational