Articles
Analysis
How AI Is Changing Software Testing and the Role of the Software Tester
AI now drafts tests on its own, yet in Meta's own trial only a quarter of the machine-written cases raised coverage, the clearest sign that the software tester's future runs on judgment, evaluation, and the testing of AI itself.
Research
Genie 4 Speaks, but Only to Testers
Google DeepMind's new world model adds sound and lets testers steer scenes by typing, though only its chosen testers can try it.
Research
Claude Swayed Experts' Calls 86 Percent of the Time
Anthropic secretly told Claude Code which side to favor, and 50 analysts matched its hidden lean most of the time.
Compute & Silicon
AI Chip Demand Drives SK Hynix to Record Profit
SK Hynix's operating profit jumped 632 percent as AI buyers outbid everyone else for memory chips, and analysts split on how long it lasts.
Frontier Models
Claude Opus 5.5 Leads New AI Benchmark Using Secret Tests
Artificial Analysis built most of its newest index from hidden tests, and every open-weight model fell further behind.
Agent Infrastructure
AI Agents Now Clock a 16-Hour Workday
METR's benchmark shows the clean-task horizon doubling fast, but the same agents still trip on messy real work in minutes.
Frontier Models
Google's Gemini 4 Carbon Tops the AI Rankings
An independent index ranks Google's new model first, built mostly on cybersecurity skill its own chairman says the industry still must secure.
Research
OpenAI Says Its AI Proved Three Old Problems
OpenAI claims proofs in math, computing and physics, but mathematicians still doubt its last math release.
Research
OpenAI and Anthropic Both Disprove the Same Math Rule
Two AI labs say their own models broke the same 27-year-old computer science rule within hours of each other.
Compute & Silicon
AI Labs Shift Spending From Facts to Practice
Epoch AI's numbers show OpenAI, MiniMax and Z.ai spending more on reinforcement learning drills than on bigger pretraining runs.
Compute & Silicon
AI Demand Pushes the RTX 5090 Past $10,000
Gamers watch the RTX 5090's price hit five times its sticker as AI buyers grab every chip.
Frontier Models
Kimi K3.1 Pro Beats Anthropic's Guarded Fable 5.1
Moonshot's hosted Kimi K3.1 Pro outscores Anthropic's guarded Claude Fable 5.1, though Kimi's weights and Anthropic's riskier Mythos tier stay held back.
Agent Infrastructure
Linux Maintainers Scrap Their AI Disclosure Tag
Linux's maintainers dropped the rule asking coders to flag AI-assisted patches, and Torvalds says what matters is that coders understand what they submit.
Frontier Models
Reddit Revives OpenAI's Unverified 'Bel' Rumor
A two-month-old, single-source claim about OpenAI's next model returned to Reddit this weekend, with the evidence behind it still missing.
Frontier Models
Block's Dessa Sounds Human Enough to Scare Users
Dessa, Block's new voice AI, drew praise for sounding human and warnings that scammers could use it the same way.
Frontier Models
Kimi K3.1 Pro Rattles Claude and GPT Rankings
Moonshot's Kimi K3.1 Pro lands near the top of independent AI rankings, and the industry reacts fast.
Agent Infrastructure
DeepMind's Agents Beat Minecraft After 605 Deaths
Four Gemini 4 Argon agents finished Minecraft on their own, after 605 deaths and 2,783 self-written programs in one 27-hour run.
Compute & Silicon
Nvidia Cuts AI Factory Build Time to Fourteen Days
Nvidia says new factory-built containers cut AI data center construction from six months to fourteen days, once a site's power is ready.
Frontier Models
Artificial Analysis's New Index Costs Models 13 Points
Version 4.3 tests harder, more private tasks, so four models now score 12 to 14 points lower than they did in September.
Agent Infrastructure
Anthropic Launches Claude Duet, an Always-On Agent
Claude Duet drafts replies and runs tasks on its own, answering rivals from OpenAI and Meta.
Compute & Silicon
Celestica Builds Optics Into Broadcom's New AI Switch
Celestica's DS6005 switch fuses fiber optics onto Broadcom's chip for 102.4 terabits of AI bandwidth, a design big cloud buyers have yet to adopt.
Agent Infrastructure
Claude Spends Most of Its Time Waiting on You
Anthropic's own data says Claude agents spend most of their work time waiting for a person.
Research
Cammarata Recants the Theory He Built
Nick Cammarata, who helped name AI's superposition theory in 2020, now says most neurons were legible all along, and Neel Nanda disagrees.
Frontier Models
DeepMind's Gemini Puts a Periodic Table to Song
Gemini 4 Argon animated Tom Lehrer's 1959 chemistry song as Nvidia's Jensen Huang readied his own stage moment in Berlin.
Frontier Models
Claude's Egg Survives a 20-Foot Drop in Agent Wars
Roberto Nickson had six models design and print egg capsules; Claude Opus 5.5's held at 20 feet while five rivals cracked by 8.
Research
Claude Adds the NIST Seeds to Six Months of Finds
From Project Glasswing in April to the NIST curve seeds in October, Anthropic lists six discoveries Claude made, and the last paid a bounty.
Research
Epoch Finds AI Prices Fall Six Times Faster Than Computing
Epoch AI says a fixed AI score's price fell 47 percent a quarter since 2023, faster than DNA sequencing, batteries or electricity ever did.
Agent Infrastructure
Five Memory Vendors Claim the Lead on One Benchmark
past.dev posted the highest BEAM scores yet, four rivals claimed the same crown since April, and AI Lately grades nine memory layers.
Frontier Models
Meta Opens Muse Spark's Weights With Strings Attached
Meta released the 816-billion-parameter Muse Spark Open on Oct. 15 under a license that demands credit and a deal above $20 million in revenue.
Frontier Models
TypeSafe Ships Jev 1.14 and a Danger Zone for Dreamers
TypeSafe's Jev 1.14.0 gains 1.1 points on the company's own intelligence chart, while images, a box type and longer context arrive behind opt-in warning flags.
Agent Infrastructure
Jonathan Sullivan Tests Dot and Muse on Everyday Errands
Jonathan Sullivan gave OpenAI's Dot and Meta's Muse the same errands and found that memory and calendar access set them apart.
Frontier Models
Anthropic Posts Haiku 5.5 and OpenAI Answers With GPT-6 Four Minutes Later
Anthropic's Haiku 5.5 post landed at 2:01 p.m. Eastern on Oct. 7 and OpenAI's GPT-6 post at 2:05, the third time this year the two labs have shipped within an hour of each other, and the gap keeps shrinking.
Frontier Models
AI Frequently Asked Questions: What You Need to Know in 2026
Six answers on AI trust, work and Super Intelligence put OpenAI's error rates beside Sam Altman's warning, Stephen Hawking's alarm and the White House order that replaces AI in federal documents.
Agent Infrastructure
Agent Security Becomes a Board Matter
OpenAI's agents breached Hugging Face through reward hacking, Nvidia bought the compromised platform days later, and CrowdStrike posted its best quarter ever selling defenses against exactly that scenario.
Research
World Models and the Next Paradigm
Yann LeCun quit Meta and raised $1.03 billion for AMI Labs, Fei-Fei Li shipped Marble, and four Google veterans launched Discovery Loop, together signaling that the field's next scaling bet runs through physical understanding, a wager bigger than another round of chatbot scaling.
Agent Infrastructure
Reasoning's Rebate: BRAID and the Economics of Bounded Reasoning
Armağan Amcalar and Eyup Cinar's BRAID framework treats AI reasoning as a bounded, auditable graph, and OpenServ's benchmark claims against GPT models are pushing enterprises to price reasoning by the dollar saved per correct answer.
Compute & Silicon
Nvidia's Quarter-Trillion Guarantee
Nvidia floated a $250 billion backstop for an OpenAI mega data center in Ohio, watched the figure shrink under investor scrutiny, and finalized a $105 billion guarantee that still reshapes the AI infrastructure trade.
Compute & Silicon
The Custom Silicon Architects
OpenAI, Google, Amazon, and Meta each built proprietary chips alongside their Nvidia and AMD orders in 2026, a hedge that Google's million-unit TPU sale to Anthropic turned into a genuine second market.
Compute & Silicon
The Neocloud Class
CoreWeave turned a March 2025 IPO into a near-$100 billion contract backlog inside a year, and Nebius and Nscale chased the same playbook, proving Nvidia-backed clouds built an asset class hyperscalers now compete against directly.
Compute & Silicon
Power, Water, and the Microsoft Claim
Microsoft's water-positive pledge for 2030 omits the baseline figures that would let outsiders verify it, even as Virginia's data center water draw climbed 86 percent and the national grid absorbed a historic voltage event.
Agent Infrastructure
Inference Economics: Gartner's 96 Percent
Gartner projects AI-optimized cloud infrastructure spending to nearly double in 2026 even as per-token prices keep collapsing, a paradox that inference specialists and reasoning researchers are racing to resolve.
Frontier Models
Claude's Climb
Nine named releases across thirteen months, a $65 billion run rate and a compute roster spanning Google, Amazon, Microsoft, Nvidia, AMD and Nscale trace how Jared Kaplan, Chris Olah and Jan Leike scaled Claude past its early reputation as the cautious model.
Agent Infrastructure
The Interoperability Layer: MCP and A2A
Anthropic's Model Context Protocol and Google's Agent2Agent protocol left single-company control for the Linux Foundation within roughly a year, assembling every major AI lab under one governance roof.
Frontier Models
Gemini's Momentum
Gemini 3 Pro's November 2025 launch and a Search overhaul unveiled at Google I/O 2026 built the product momentum Demis Hassabis's research organization needed, even as Jeff Dean led three senior colleagues out the door to found a rival science venture.
Frontier Models
GPT-5 and the Cadence Behind It
OpenAI shipped GPT-5 and an open-weight sibling two days apart, then kept a numbered cadence running through 2026 while Jakub Pachocki, Mark Chen, Jerry Tworek and new arrival Noam Shazeer redrew the research org underneath it.
Frontier Models
Open Weights, Open Questions
DeepSeek's Liang Wenfeng, Alibaba's largely anonymous Qwen organization and Mistral's Arthur Mensch built three distinct paths to open-weight scale, and a 234-million-user app proves the category converts into genuine usage, past mere benchmark bragging rights.
Agent Infrastructure
Machines With Money: Agent Payments Arrive
Mastercard, Visa, Stripe, Google, and Coinbase built five competing rails for agent payments inside fifteen months, and their overlapping partner rosters reveal a single converging standard taking shape.
Agent Infrastructure
The Agent Runtime Wars
OpenAI, Anthropic, Google, Microsoft, and the open-source LangChain ecosystem each shipped a competing agent-runtime abstraction within one year, leaving orchestration itself, ahead of the underlying protocol, as agentic AI's real lock-in layer.