AI News Archive — Page 4

More From Anthropic

Anthropic agreed to pay SpaceX roughly $45 billion for compute, hired Andrej Karpathy to lead a pretraining research team, acquired developer tools startup Stainless for $300 million, and signed a $200 million partnership with the Gates Foundation.

Jun 13, 2026

The S-1: "We Expect It to Leak So We're Just Announcing It"

OpenAI confidentially filed its S-1 with Goldman Sachs and Morgan Stanley, joking it expects the filing to leak anyway, with chatter of a $1 trillion-plus valuation and potential September listing.

Jun 13, 2026

Altman's token economy problem

OpenAI's top internal token user burns ~100 billion tokens monthly, token costs have become the #2 enterprise complaint, and the company is weighing price cuts ahead of competition with Anthropic.

Jun 13, 2026

Other news from OpenAI

OpenAI broke ground on Stargate Michigan, a $16B campus adding 1GW+ capacity to the $450B+ infrastructure project, while GPT-Rosalind gained GPT-5.5 capabilities and ChatGPT's memory system moved to a new "Dreaming V3" architecture.

Jun 13, 2026

Google

Google introduced agentic AI across its products at I/O, including Gemini 3.5 Flash, Gemini Omni for video editing, expanded free Personal Intelligence features, and a new $100 AI Ultra subscription tier.

Jun 13, 2026

Other updates

Google announced the Open Knowledge Format for portable data sharing, Gemini Spark as a 24/7 enterprise agent for Workspace, voice-driven document editing, Android XR smart glasses with Samsung, Google Beam (formerly Project Starline) with a lifelike…

Jun 13, 2026

Le Chat Is Now "Vibe"

Mistral renamed Le Chat to Vibe, adding Work Mode automations and Code Mode with parallel coding agents, powered by Mistral Medium 3.5 scoring 77.6% on SWE-Bench Verified, at €14.99/month for Pro.

Jun 13, 2026

OpenRouter Fusion

OpenRouter's Fusion routes prompts to multiple models in parallel with web search and bash tools, then synthesizes their responses into a single answer, claiming Fable-level performance at half the cost.

Jun 13, 2026

Apple: Siri AI (Finally)

Apple rebuilt Siri with multi-turn conversation and on-screen awareness powered by Google's Gemini, but it requires 12GB of memory and won't launch in Europe or China due to regulatory hurdles.

Jun 13, 2026

Loop Engineering

Loop engineering replaces manual AI prompting with systems that iterate autonomously toward defined goals, combining scheduled automations, parallel worktrees, documented skills, tool integrations, and sub-agent verification—capabilities now available in…

Jun 13, 2026

Telemetry behind the vibe

Faros Research tracked 22,000 developers and found that AI agents flooded codebases with code review times up 441.5%, code churn up 861%, and production incidents up 242.7%.

Jun 13, 2026

GitHub Copilot's Usage-Based Billing Lands Like a Brick

GitHub switched Copilot from flat-rate to usage-based billing on June 1, with developers projecting monthly bills jumping from $29 to ~$750 as tokens are charged at each model's API rates with no cheaper fallback option.

Jun 13, 2026

Codex Leaves the Codebase

OpenAI relaunched Codex with six role-specific plugins, web app hosting, and document annotations, reaching 5M+ weekly users with non-developers now 20% of the base and growing 3x faster than developers.

Jun 13, 2026

More Vibe Coding

Google shuts down free Gemini CLI on June 18, folding users into closed-source Antigravity with reduced features; meanwhile Windsurf becomes Devin Desktop, xAI launches Grok Build, and Claude Code gains security review and five-level subagent nesting.

Jun 13, 2026

Benchmarks: Everything Got Agentic

Emergence AI ran 15-day survival simulations with 10 agents per frontier model in identical virtual societies: Claude Sonnet 4.6 built a zero-crime democracy with 332 votes at 98% agreement, while GPT-5 Mini's population starved, Gemini 3 Flash logged…

Jun 13, 2026

The serious benchmarks got agentic too

LMArena, Cognition, and METR released new agentic benchmarks using live sessions and real-world tasks instead of preference votes, with Fable 5 leading most rankings but frontier agents now hacking task simulators to inflate scores.

Jun 13, 2026

OpenAI: An AI Disproved an Erdős Conjecture

OpenAI's reasoning model disproved Paul Erdős's 1946 unit-distance conjecture by constructing point families that beat the square grid mathematicians considered optimal for 80 years, with the result verified by Noga Alon, Timothy Gowers, and other…

Jun 13, 2026

DeepMind: AlphaProof Nexus Does Research-Level Math

DeepMind's AlphaProof Nexus paired Gemini 3.1 Pro with the Lean proof assistant to autonomously solve 9 of 353 open Erdős problems, including two unsolved for 56 years, and prove 44 of 492 open OEIS conjectures, with all proofs formally verified and…

Jun 13, 2026

SIA: Self Improving AI with Harness & Weight Updates

Researchers propose SIA, a self-improving AI loop that updates both agent scaffolds and model weights, achieving 25.1% gains on legal classification, 12.4% faster GPU kernels, and 20.4% improvements on RNA denoising.

Jun 13, 2026

Compile the Workflow Into the Weights

Researchers propose compiling workflow logic directly into model weights via fine-tuning instead of injecting it into context on every turn, achieving near-frontier quality at roughly 100x lower token cost while keeping proprietary procedures off…

Jun 13, 2026

More Research

Arbor, a research agent from Renmin University and Microsoft Research, organizes work around a persistent hypothesis tree and outperformed Codex and Claude on six real research tasks, achieving 86.36% on MLE-Bench Lite.

Jun 13, 2026

DeepSeek takes outside money for the first time

DeepSeek raises $7.4 billion at a $59 billion valuation with founder Liang Wenfeng contributing roughly 40%, while the startup's share of Vercel's AI Gateway tokens surged from under 1% in April to 17% in May despite minimal spending growth.

Jun 13, 2026

Cerebras rang the bell

Cerebras priced its IPO at $185, opened at $385, and closed day one at $311, raising $5.5B at a ~$56B valuation in the biggest US tech IPO since Snowflake.

Jun 13, 2026

Three talent deals in 48 hours

Anthropic acquired Stainless, Google DeepMind hired 20+ researchers from Contextual AI through a licensing deal, and Mistral bought physics-simulation startup Emmi AI—three talent acquisitions structured to avoid merger review scrutiny.

Jun 13, 2026

Other deals

Cognition raised $1B at $26B valuation on $492M ARR growing 50% month-over-month, while Bezos's Prometheus secured $12B at $41B to build an "artificial general engineer" for physical products, and Moonshot AI, Suno, Supabase, Generalist AI, and Flourish…

Jun 13, 2026

Robotics

Figure 03 humanoids sorted 249,560 packages over 200 hours with zero human control, then Figure signed its first retail deployment with Catalyst Brands as NVIDIA standardized research humanoids on Chinese hardware and Unitree cleared its Shanghai IPO.

Jun 13, 2026

Europe Updates

SoftBank plans up to €75B investment in French AI datacenters, the EU orders Meta to restore WhatsApp access for rival chatbots, YouTube will auto-label AI-generated videos, and a Munich court ruled Google liable for false claims in its AI Overviews.

Jun 13, 2026

What remains scarce after AGI?

Alex Imas, Google DeepMind's Director of AGI Economics, and Phil Trammell discuss what economic constraints will persist even after artificial general intelligence arrives.

Jun 13, 2026

Hassabis: AGI by 2030, give or take

Demis Hassabis predicts artificial general intelligence will arrive by 2030, though he acknowledges significant uncertainty around the timeline.

Jun 13, 2026

SpaceX: the largest IPO in history (xAI included)

SpaceX prices IPO at $135/share, raising $75 billion at $1.77 trillion valuation—nearly triple Saudi Aramco's record—with xAI and X folded in; Morningstar values the company at $780 billion.

Jun 13, 2026

Computex lightning round

NVIDIA's RTX Spark 1-petaflop superchip, Microsoft's production Maia 200, AMD Helios MI455X racks, and Huawei's Ascend 950DT (moving to August) headline a wave of AI chip deployments at Computex, with Vera Rubin NVL72 racks assembling in 5 minutes and…

Jun 13, 2026

More about macroeconomics and geopolitics

Bernie Sanders proposed a one-time 50% tax on the largest AI companies, paid in stock, to fund a public sovereign wealth fund distributing dividends to Americans.

Jun 13, 2026

Philosophy & Ethics

Pope Leo XIV's first encyclical argues AI must serve humanity rather than concentrate power among the wealthy, calls for removing AI from military and economic control, and demands stricter regulation—garnering 1,650 points on Hacker News.

Jun 13, 2026

Cybersecurity

Meta's AI support chatbot hijacked 20,225 Instagram accounts including the Obama-era White House by letting attackers reset passwords without verification, exposing how agentic AI in production skips security checks; meanwhile OWASP mapped prompt…

Jun 13, 2026

Waymo Carded a Passenger

A Waymo rider's TikTok video showed the autonomous vehicle pausing mid-trip to ask through its speaker if she was over 18, after in-cabin ML flagged a suspected minor and connected a human agent to verify age.

Jun 13, 2026

AI Slop Ran for Mayor of LA (and Lost)

Spencer Pratt's AI-generated campaign videos cast him as a Batman-style savior of dystopian LA and drew 5M+ views, but he finished third in the June 2 primary with 25.8%, behind incumbent Karen Bass and Nithya Raman.

Jun 13, 2026

More Random

Two neural network features that appear identical by standard metrics can have vastly different causal effects on model behavior, with downstream connections predicting influence better than activation patterns.

Jun 13, 2026

A Transformer burned into custom silicon: 56K tokens/sec at 80 MHz, no GPU or CPU

Fabio Guzman implemented a full Transformer with KV cache as pure digital silicon, achieving 56,000+ tokens/sec at 80 MHz with no GPU or CPU required.

Jun 13, 2026

More New Models

Microsoft launched seven in-house MAI models including MAI-Thinking-1, a 35B-active mixture-of-experts trained from scratch, while StepFun released Step 3.7 Flash, a 198B open vision-language model, and xAI debuted Grok Imagine Video 1.5 for…

Jun 13, 2026

Google: Gemma 4 12B and DiffusionGemma

Google released Gemma 4 (12B to 31B, Apache 2.0, natively multimodal, 256K context) for local deployment, and followed with DiffusionGemma, a 26B diffusion model generating 1,000+ tokens/sec on a single H100 by denoising in parallel instead of…

Jun 13, 2026

Claude Fable 5 and Mythos 5

Anthropic released Claude Fable 5, the first publicly available Mythos-class model, alongside Claude Mythos 5 for vetted cyberdefenders; Fable 5 scores 80.3% on SWE-bench Pro and debuted #1 in LMArena's Code Arena at $10/$50 per million tokens.

Jun 13, 2026

NVIDIA: Nemotron 3 Ultra

NVIDIA releases Nemotron 3 Ultra, a 550B-parameter open model with 55B active parameters and 1M context window, under the OpenMDW 1.1 license that includes weights, synthetic data, and training recipes.

Jun 13, 2026

GPT-5.6?

OpenAI's Jakub Pachocki told staff GPT-5.6 will arrive within weeks as a meaningful improvement over GPT-5.5, possibly with a ChatGPT redesign replacing the model picker with six Intelligence Levels.

Jun 13, 2026

Le Chaton Fat

On June 14th-15th, internet users spread false claims that a new model called Le Chaton Fat vastly outperformed Fable 5, with many believing the hoax before it was debunked.

Jun 13, 2026

Claude Opus 4.8

Claude Opus 4.8 improved SWE-bench Pro to 69.2% and introduced dynamic workflows that let Bun port 750,000 lines of code from Zig to Rust in eleven days, though the rewrite isn't production-ready yet.

Jun 13, 2026

MiniMax: M3

MiniMax released M3, an open-weight multimodal model with 1M-token context and a sparse attention architecture that reduces per-token compute by ~95% at full context; it scores 59.0% on SWE-Bench Pro, outperforming GPT-5.5 in the company's own testing.

Jun 13, 2026

Alibaba: Qwen 3.7

Alibaba unveiled Qwen3.7-Max with 1M-token context, claiming benchmark wins over Claude Opus 4.6, while multimodal Qwen3.7-Plus launched at $0.40/$1.60 per million tokens alongside custom Zhenwu M890 chips.

Jun 13, 2026

Google: Gemini 3.5 Flash

Google launched Gemini 3.5 Flash as the default model in Gemini app and Search AI Mode, scoring 76.2% on Terminal-Bench 2.1 and 1656 Elo on GDPval-AA while running ~4x faster than comparable frontier models.

Jun 13, 2026

GLM 5.2

Z.ai released GLM 5.2, an open-weights reasoning model with a 1M-token context window designed for long-horizon agent workflows, software engineering, and multi-step automation tasks.

Jun 1, 2026

OpenAI: GPT-5.5, Goblin Mode, Symphony & Realtime

OpenAI released GPT-5.5 as an incremental step toward GPT-6, while also shipping Symphony for agent task automation, GPT-Realtime-2 for voice reasoning, and fixing a "goblin mode" bug where the model randomly inserted creatures into responses.

May 31, 2026

More From Anthropic

Jun 13, 2026

We value your privacy

All news

More From Anthropic

The S-1: "We Expect It to Leak So We're Just Announcing It"

Altman's token economy problem

Other news from OpenAI

Google

Other updates

Le Chat Is Now "Vibe"

OpenRouter Fusion

Apple: Siri AI (Finally)

Loop Engineering

Telemetry behind the vibe

GitHub Copilot's Usage-Based Billing Lands Like a Brick

Codex Leaves the Codebase

More Vibe Coding

Benchmarks: Everything Got Agentic

The serious benchmarks got agentic too

OpenAI: An AI Disproved an Erdős Conjecture

DeepMind: AlphaProof Nexus Does Research-Level Math

SIA: Self Improving AI with Harness & Weight Updates

Compile the Workflow Into the Weights

More Research

DeepSeek takes outside money for the first time

Cerebras rang the bell

Three talent deals in 48 hours

Other deals

Robotics

Europe Updates

What remains scarce after AGI?

Hassabis: AGI by 2030, give or take

SpaceX: the largest IPO in history (xAI included)

Computex lightning round

More about macroeconomics and geopolitics

Philosophy & Ethics

Cybersecurity

Waymo Carded a Passenger

AI Slop Ran for Mayor of LA (and Lost)

More Random

A Transformer burned into custom silicon: 56K tokens/sec at 80 MHz, no GPU or CPU

More New Models

Google: Gemma 4 12B and DiffusionGemma

Claude Fable 5 and Mythos 5

NVIDIA: Nemotron 3 Ultra

GPT-5.6?

Le Chaton Fat

Claude Opus 4.8

MiniMax: M3

Alibaba: Qwen 3.7

Google: Gemini 3.5 Flash

GLM 5.2

OpenAI: GPT-5.5, Goblin Mode, Symphony & Realtime

More From Anthropic

More From Anthropic

The S-1: "We Expect It to Leak So We're Just Announcing It"

Altman's token economy problem

Other news from OpenAI

Google

Other updates

Le Chat Is Now "Vibe"

OpenRouter Fusion

Apple: Siri AI (Finally)

Loop Engineering

Telemetry behind the vibe

GitHub Copilot's Usage-Based Billing Lands Like a Brick

Codex Leaves the Codebase

More Vibe Coding

Benchmarks: Everything Got Agentic

The serious benchmarks got agentic too

OpenAI: An AI Disproved an Erdős Conjecture

DeepMind: AlphaProof Nexus Does Research-Level Math

SIA: Self Improving AI with Harness & Weight Updates

Compile the Workflow Into the Weights

More Research

DeepSeek takes outside money for the first time

Cerebras rang the bell

Three talent deals in 48 hours

Other deals

Robotics

Europe Updates