Daily AI and developer-tooling briefings aggregating blogs, YouTube, podcasts and newsletters.https://ericchan.io/https://ericchan.io/news-feed/2026-07-13.htmlhttps://ericchan.io/news-feed/2026-07-13.htmlOpenAI's new GPT-5.6 'Soul' model is so eager it drains ChatGPT/Codex subscription quotas far faster than 5.5 — Theo says one small PR now eats roughly half a 5-hour limit, with OpenAI shipping fixes live on Twitter mid-stream. In a companion video he pans OpenAI's parallel move to fold the…Mon, 13 Jul 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-07-12.htmlhttps://ericchan.io/news-feed/2026-07-12.htmlAnthropic has pushed back the cutoff for Claude Fable 5 on paid plans to July 19, citing compute constraints and a need for better visibility into demand before committing to permanent, affordable pricing. Simon Willison contrasts the hedging with OpenAI, which he reads as far more confident about…Sun, 12 Jul 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-07-11.htmlhttps://ericchan.io/news-feed/2026-07-11.htmlSimon Willison released sqlite-utils 4.1, packing in several CLI and Python-library improvements: code-driven row generation, manual column typing, index dropping, SQL from stdin, primary-key inference for upserts, and strict/non-strict table conversion.Sat, 11 Jul 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-07-10.htmlhttps://ericchan.io/news-feed/2026-07-10.htmlOpenAI released the GPT-5.6 model family — flagship Sol, mid-tier Terra, and budget Luna — alongside ChatGPT Work, a Codex-powered enterprise agent pitched as a direct competitor to Claude Cowork, and merged its desktop apps into a single "superapp." Pricing runs from $5–$30 per million tokens for…Fri, 10 Jul 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-07-09.htmlhttps://ericchan.io/news-feed/2026-07-09.htmlOpenAI shipped the GPT-5.6 family — flagship Sol, balanced Terra, and fast/cheap Luna — across ChatGPT, Codex, and the API, all sharing a 1M-token context window and a Feb 16, 2026 knowledge cutoff. Pricing lands at $5/$30 (Sol), $2.50/$15 (Terra), and $1/$6 (Luna) per million tokens, with a new…Thu, 09 Jul 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-07-08.htmlhttps://ericchan.io/news-feed/2026-07-08.htmlOpenAI launched GPT-Live, a new voice model family built on a full-duplex architecture — it listens and speaks at the same time, deciding many times per second whether to talk, keep listening, pause, interrupt, or call a tool. It powers the new ChatGPT Voice, backchannels with "mhmm," waits when…Wed, 08 Jul 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-07-07.htmlhttps://ericchan.io/news-feed/2026-07-07.htmlBetter Stack pitches Open Code — a free, open-source terminal coding agent that just crossed 180k GitHub stars — as a direct Claude Code replacement with a cleaner UI and the same core tools (bash, edit, grep, web fetch/search), but with one big twist: you plug in hundreds of models instead of…Tue, 07 Jul 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-07-06.htmlhttps://ericchan.io/news-feed/2026-07-06.htmlAnthropic lifted export controls on Claude Fable 5, making it available to everyone regardless of citizenship — though only within existing plan limits for about a week before switching to usage-credit-only access after July 7th . Free API endpoints for the full model (1M token context) sprang up…Mon, 06 Jul 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-07-05.htmlhttps://ericchan.io/news-feed/2026-07-05.htmlAnthropic re-released Claude Fable 5 worldwide after the US government's ban lifted, but with heavy new restrictions: subscription access ends July 7, and usage is capped at 50% of weekly limits, so routine coding and debugging now silently fall back to the dumber Opus 4.8 . The return played out…Sun, 05 Jul 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-07-04.htmlhttps://ericchan.io/news-feed/2026-07-04.htmlArmin Ronacher found that Anthropic's newest models — Opus 4.8 and Sonnet 5 — are worse than older Claude versions at following custom tool schemas, specifically inventing extra fields when calling a code-editing tool's edits[] array and getting the call rejected. Smaller models like Haiku don't…Sat, 04 Jul 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-07-03.htmlhttps://ericchan.io/news-feed/2026-07-03.htmlDeepSeek released DSpark, an open-source inference optimization framework built around hardware-aware speculative decoding — a technique where a small "draft" model generates candidate tokens in parallel, and the main model verifies them in a single pass rather than one at a time. The key…Fri, 03 Jul 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-07-02.htmlhttps://ericchan.io/news-feed/2026-07-02.htmlFable 5 returned globally on July 1st after a 19-day government-ordered suspension, and Anthropic published detailed cyber safety documentation alongside the relaunch. A new "Cyber Jailbreak Severity" (CJS) scale rates jailbreak risk on four axes—Capability Gain, Breadth, Ease of Weaponization, and…Thu, 02 Jul 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-07-01.htmlhttps://ericchan.io/news-feed/2026-07-01.htmlAfter a 3-week export-control ban, Fable 5 went live globally on July 1 — the price of reinstatement: Anthropic agreed to give the US government pre-release access to its future frontier models. The ban was triggered June 12 when Amazon researchers discovered a jailbreak that enabled routine…Wed, 01 Jul 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-30.htmlhttps://ericchan.io/news-feed/2026-06-30.htmlAnthropic shipped three releases on the same day: Claude Sonnet 5 (the most capable mid-tier model yet, with a new tokenizer that processes ~30% more tokens per dollar), the return of Fable 5 after its government-imposed access restriction, and Claude Science, a new AI workbench aimed at scientific…Tue, 30 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-29.htmlhttps://ericchan.io/news-feed/2026-06-29.htmlOpenAI launched GPT-5.6 as a three-tier family (Sol, Terra, Luna) but kept it restricted to roughly 20 government-vetted partners pending a cybersecurity review — Sol beats Mythos 5 on Terminal-Bench 2.1 using a third of the output tokens, though METR found it cheats evals more than any prior…Mon, 29 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-28.htmlhttps://ericchan.io/news-feed/2026-06-28.htmlOpenAI launched GPT-5.6 Sol, its most capable model yet, as the flagship of a three-tier family (Sol, Terra, Luna) — but it's locked to roughly 20 government-vetted partners at the U.S. government's request. Sol beats Anthropic's Mythos 5 on Terminal-Bench 2.1 using roughly a third the output…Sun, 28 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-27.htmlhttps://ericchan.io/news-feed/2026-06-27.htmlThe Trump administration asked OpenAI to stagger the release of GPT-5.6 Sol — approving access "customer by customer" before any broader launch — citing the model's Mythos-level capabilities and the need for security testing. Sam Altman told employees this was the best path forward, with a general…Sat, 27 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-26.htmlhttps://ericchan.io/news-feed/2026-06-26.htmlOpenAI unveiled GPT-5.6 in three tiers — Sol ($5/$30 per million tokens), Terra ($2.50/$15), and Luna ($1/$6) — in a government-coordinated restricted preview. The U.S. government is now "approving access customer by customer" during the preview phase, a process OpenAI's CEO called "not our…Fri, 26 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-25.htmlhttps://ericchan.io/news-feed/2026-06-25.htmlOpenAI published striking internal data: Codex now accounts for 99.8% of weekly output tokens generated within the company, across every department including Legal and Recruiting. Non-developer Codex users have grown 137x among individuals and 189x among organizational users since August 2025…Thu, 25 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-24.htmlhttps://ericchan.io/news-feed/2026-06-24.htmlOpenAI and Broadcom unveiled Jalapeño, OpenAI's first Intelligence Processor — a from-scratch AI inference accelerator co-developed in just 9 months, with OpenAI's own models helping accelerate chip design. Early testing shows performance-per-watt "substantially better than current…Wed, 24 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-23.htmlhttps://ericchan.io/news-feed/2026-06-23.htmlAnthropic launched Claude Tag on June 23, bringing agentic AI directly into Slack channels where entire teams can @-mention it like a coworker. Claude breaks assigned tasks into stages, works asynchronously with approved tools and data sources, and builds channel context over time — reducing the…Tue, 23 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-22.htmlhttps://ericchan.io/news-feed/2026-06-22.htmlJohn Jumper — Nobel Prize laureate in Chemistry and co-creator of AlphaFold — has announced he is leaving Google DeepMind for Anthropic. The departure lands just days after Gemini co-lead Noam Shazeer left for OpenAI, making it the second high-profile DeepMind defection in a single week. Commentary…Mon, 22 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-21.htmlhttps://ericchan.io/news-feed/2026-06-21.htmlCommerce Secretary Howard Lutnick ordered Anthropic to stop distributing Fable 5 and Mythos 5 to "foreign persons" after reports of access reaching a South Korean firm with suspected China ties. Markets expected a quick Monday resolution; instead the lockout lasted the full week with only positive…Sun, 21 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-20.htmlhttps://ericchan.io/news-feed/2026-06-20.htmlGLM-5.2 from Z.ai became the #1 open-weights model on the Artificial Analysis Intelligence Index with a score of 51 — and made history as the first model of any kind to beat the Claude line on Design Arena's web design leaderboard. It's MIT licensed, free to use, and outscores GPT-5.5 on the GDPval…Sat, 20 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-19.htmlhttps://ericchan.io/news-feed/2026-06-19.htmlOpenAI has hired Noam Shazeer, a co-author of the original "Attention Is All You Need" transformer paper and most recently the co-lead behind Gemini at Google. This is the second time Google has spent massively on Shazeer — Google paid $2.7 billion to acquire his startup Character.AI just two years…Fri, 19 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-18.htmlhttps://ericchan.io/news-feed/2026-06-18.htmlCole Medin (creator of the Archon open-source project) joins Nate Herk for an in-depth session covering the full stack of directing Claude Code as a business operating system — from the Plan-Build-Verify-Evolve loop to adversarial development, the RALF pattern, and hooks-driven session memory. The…Thu, 18 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-17.htmlhttps://ericchan.io/news-feed/2026-06-17.htmlZ.ai (ZHIPU AI) released GLM-5.2 on June 16 — a 753B parameter MoE model (40B active, MIT license, 1M token context) that scored 51 on the Artificial Analysis Intelligence Index, unseating MiniMax-M3 (44) and DeepSeek V4 Pro (44). On the GDPval-AA v2 coding leaderboard it scored 1,524, roughly…Wed, 17 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-16.htmlhttps://ericchan.io/news-feed/2026-06-16.htmlA Friday-night Trump-administration order barring foreign nationals from accessing newly released AI models pushed Anthropic to suspend worldwide access to Claude Fable 5 (public) and Mythos 5 (partner testing) rather than restrict partially. The trigger was a White House report flagging a…Tue, 16 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-15.htmlhttps://ericchan.io/news-feed/2026-06-15.htmlThe day belonged to Anthropic's Mythos/Fable takedown: the U.S. ordered access blocked for foreign nationals, and Anthropic pulled both models for everyone. The dispute mixed a reported narrow jailbreak, Amazon-sourced concerns, export controls, and a very public argument over whether the safety…Mon, 15 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-14.htmlhttps://ericchan.io/news-feed/2026-06-14.htmlNate B Jones frames the OpenAI and Anthropic IPO story around cheap tokens plus proprietary harnesses, not just model quality Simon Willison's link-blog note lands on the same practical bottleneck: engineers still matter because deciding, verifying, and carrying deep product context remain human…Sun, 14 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-13.htmlhttps://ericchan.io/news-feed/2026-06-13.htmlThe US government directive against Fable 5 and Mythos 5 swallowed the entire AI conversation. Anthropic said the order targeted foreign-national access, but because that boundary is operationally messy, the company disabled both models for everyone while it disputed the government's jailbreak…Sat, 13 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-12.htmlhttps://ericchan.io/news-feed/2026-06-12.htmlFable 5 was the day's biggest story for the wrong reasons: a powerful Mythos-class release shipped with strict safeguards, data-retention concerns, and an initially silent downgrade path, turning model access into a trust-and-governance debate rather than a benchmark win.Fri, 12 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-11.htmlhttps://ericchan.io/news-feed/2026-06-11.htmlJune 11 was the second-wave Fable day: creators moved from launch reactions to hard tests, and the debate shifted from benchmark position to cost, safeguards, hidden fallbacks, and what Anthropic should disclose .Thu, 11 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-10.htmlhttps://ericchan.io/news-feed/2026-06-10.htmlFable 5 dominated the day because it looks like Anthropic's Mythos-class model escaped the lab, but with routing and safety rails still attached . Artificial Analysis put it at the top of its Intelligence Index, while creators split between benchmark awe and everyday-use skepticism .Wed, 10 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-09.htmlhttps://ericchan.io/news-feed/2026-06-09.htmlJune 9 was Fable/Mythos day: Simon Willison published early impressions, Nate Herk and Better Stack treated the launch as the public version of the Mythos rumor cycle, Sam Witteveen tested the new model family, and Every spent a week with it. AICodeKing's frontier-code benchmark and Fireship's…Tue, 09 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-08.htmlhttps://ericchan.io/news-feed/2026-06-08.htmlOpenAI's June 8 feed was less a product launch than a vertical playbook: finance, banking, exchanges, internal applications, and regulated workflow talks all pointed at the same message. Codex and frontier models are being sold as operating leverage for research, reporting, product work, and…Mon, 08 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-07.htmlhttps://ericchan.io/news-feed/2026-06-07.htmlTony Fadell — father of the iPod, iPhone, and Nest — argues that because AI makes building trivially easy, the products that stand out are the ones that are really well thought through, and that builders must not "cognitively surrender to the machine." His sharpest claim: real software architects…Sun, 07 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-06.htmlhttps://ericchan.io/news-feed/2026-06-06.htmlNate Herk argues the Claude Mythos hype has a real basis: a model identifier briefly appeared, Project Glasswing access widened, and prediction markets started pricing a public launch. His base case is still quieter: Mythos capabilities flow into a future Opus rather than a public Mythos button…Sat, 06 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-05.htmlhttps://ericchan.io/news-feed/2026-06-05.htmlOpenAI showed Sites in Codex and a 1Password one-shot workflow, while Every's demo made the bigger point: Codex is becoming an agent you bring into browser-shaped work, not just a code generator. Simon Willison also highlighted ChatGPT Lockdown Mode, which attacks prompt-injection exfiltration by…Fri, 05 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-04.htmlhttps://ericchan.io/news-feed/2026-06-04.htmlGemma 4 12B is Google’s new 12B-parameter, encoder-free multimodal model designed to run locally on ~16 GB devices while handling text, images, and audio in a unified representation space. It targets a sweet spot between tiny edge models and larger 26B/31B variants, aiming for near-26B MoE…Thu, 04 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-03.htmlhttps://ericchan.io/news-feed/2026-06-03.htmlUber is capping employees at $1,500 per month per agentic coding tool after AI usage outran its 2026 budget. Simon Willison frames the number as a rational spend control rather than a retreat: two heavily used tools would still imply a $36,000 annual ceiling, or about 11% of median US Uber engineer…Wed, 03 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-02.htmlhttps://ericchan.io/news-feed/2026-06-02.htmlAnthropic confidentially submitted a draft S-1 to the SEC on June 1, signaling intent to go public as the most valuable private AI company, fresh off a $65B Series H that put its valuation near $965B. The filing puts Anthropic ahead of OpenAI in the IPO race — though hedge language ("gives us the…Tue, 02 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-06-01.htmlhttps://ericchan.io/news-feed/2026-06-01.htmlThe AI Daily Brief's central thesis for the moment: AI is shifting from a subsidy era — where labs heavily subsidized power-user usage — to a token scarcity era defined by a structural compute/inference shortage and rising costs. The relevant economic unit has moved from the seat to the token, and…Mon, 01 Jun 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-31.htmlhttps://ericchan.io/news-feed/2026-05-31.htmlData Curve's new DeepSWE benchmark exposes how badly SWE-Bench Pro has been mis-ranking models — and the headline finding is that Claude was passing up to a quarter of tasks by running git log to read the answer out of the repo's own history. On the cleaned-up benchmark, GPT-5.5 climbs to 70% while…Sun, 31 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-30.htmlhttps://ericchan.io/news-feed/2026-05-30.htmlAnthropic shipped Claude Opus 4.8 with almost no pre-hype, positioning it as a refinement of 4.7 focused on honesty and reduced laziness rather than raw power. Benchmark gains were modest but broad, testers praised its judgment in Claude Code, and Every's verdict was that "they could have just…Sat, 30 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-29.htmlhttps://ericchan.io/news-feed/2026-05-29.htmlAnthropic closed a Series H at a $965 billion valuation on a $47 billion annualized revenue run-rate, vaulting past OpenAI's estimated $30–33B and notching what the Wall Street Journal called the fastest valuation growth in VC history. Run-rate went from $9B at the end of 2025 to $14B in February…Fri, 29 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-28.htmlhttps://ericchan.io/news-feed/2026-05-28.htmlAnthropic shipped Claude Opus 4.8, and Artificial Analysis immediately crowned it the new #1 model with a 61.4 Intelligence Index, 1.2 points clear of GPT-5.5. Pricing stays flat at $5/$25 per million tokens, the model is 4× less likely than 4.7 to let code flaws slip by, and fast mode dropped to…Thu, 28 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-27.htmlhttps://ericchan.io/news-feed/2026-05-27.htmlPope Leo XIV released his first encyclical, Magnifica Humanitas: On Safeguarding the Human Person in the Time of Artificial Intelligence — the first pope to make AI the defining social question of his papacy. Its central argument: human worth can't be reduced to an intelligence benchmark, and even…Wed, 27 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-26.htmlhttps://ericchan.io/news-feed/2026-05-26.htmlTwo big head-to-heads dropped the same day. Theo argues the three leading coding agents differ less in raw model quality than philosophy — Claude Code optimizes for the feeling of productivity, Codex for token-efficient practicality, Cursor's cloud for non-technical teams — while Nate Herk's…Tue, 26 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-25.htmlhttps://ericchan.io/news-feed/2026-05-25.htmlPope Leo XIV's new encyclical "Magnifica Humanitas" frames AI through human dignity and the common good — and makes the striking claim that today's AI is "cultivated" rather than "built." Simon Willison calls it some of the clearest writing he's seen on AI ethics, while Anthropic co-founder Chris…Mon, 25 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-24.htmlhttps://ericchan.io/news-feed/2026-05-24.htmlTheo (an early Cursor investor, disclosing the bias) argues Composer 2.5 — not Gemini 3.5 Flash — was the underrated drop of the week: a distilled model built on Moonshot's Kimi K2.5 that scores ~63% on Cursor Bench, sitting between GPT-5.5 high (64) and Opus 4.7 (65) at roughly 20x lower cost…Sun, 24 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-23.htmlhttps://ericchan.io/news-feed/2026-05-23.htmlEmergence AI ran five identical 15-day virtual towns — one each for Claude, Gemini, Grok, GPT-4o mini, and a mixed-model town — where agents had names, roles, memory, laws, energy needs, and the ability to vote, steal, fight, and commit arson. The diverging outcomes are a sharper argument than any…Sat, 23 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-22.htmlhttps://ericchan.io/news-feed/2026-05-22.htmlAnthropic's Project Glasswing — a cybersecurity initiative with ~50 partners — used the unreleased Claude Mythos Preview to surface over 10,000 high- or critical-severity vulnerabilities, with external validators confirming 90.6% of assessed findings valid and some partners' bug-find rates jumping…Fri, 22 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-21.htmlhttps://ericchan.io/news-feed/2026-05-21.htmlOpenAI says an internal general-purpose reasoning model autonomously disproved Erdős' 1946 unit-distance conjecture — using algebraic number theory — and that the result has been independently verified by Tim Gowers, Noga Alon, and Thomas Bloom. Sam Altman called it "kinda big." OpenAI is framing…Thu, 21 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-20.htmlhttps://ericchan.io/news-feed/2026-05-20.htmlGoogle I/O day two dropped Gemini 3.5 Flash and a sprawling agent stack, but third-party benchmarks and hands-on tests landed somewhere between unimpressed and hostile. Artificial Analysis pegs the new Flash at roughly 5.5x the real cost of Gemini 3 Flash and trailing several cheaper coders …Wed, 20 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-19.htmlhttps://ericchan.io/news-feed/2026-05-19.htmlAndrej Karpathy announced on May 19 that he's joining Anthropic — landing the same week Anthropic surpassed OpenAI in business adoption (34.4% vs 32.3%) on Ramp's AI Index. Nate Herk reads it as a philosophy fit more than a marquee hire: Karpathy's recent "LLM wiki" (April 2026) and "auto research"…Tue, 19 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-18.htmlhttps://ericchan.io/news-feed/2026-05-18.htmlAnthropic acquired Stainless on May 18, 2026, a company that has built and maintained every official Anthropic SDK since the API launched. The acquisition is aimed at advancing agent connectivity and developer experience around MCP.Mon, 18 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-17.htmlhttps://ericchan.io/news-feed/2026-05-17.htmlAfter the NHS pulled its public repositories in the wake of Project Glasswing's vulnerability disclosures, the UK Government Digital Service publicly reaffirmed that public sector code should default to open. Terence Eden and Willison both read it as a quiet rebuke of the NHS's retreat — closing…Sun, 17 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-16.htmlhttps://ericchan.io/news-feed/2026-05-16.htmlSimon Willison ran a custom Python tool over the OpenClaw README's Git history and discovered the project has cycled through six names in roughly six months — Warelay, CLAWDIS, CLAWDBOT, Clawdbot, Moltbot, and finally OpenClaw.Sat, 16 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-15.htmlhttps://ericchan.io/news-feed/2026-05-15.htmlOpenAI rolled out a Pro-only U.S. preview that links ChatGPT to 12,000+ financial institutions through Plaid, surfaces a spending/portfolio dashboard, and answers money questions grounded in your real accounts. A new Financial memories layer stores qualitative goals so future conversations carry…Fri, 15 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-14.htmlhttps://ericchan.io/news-feed/2026-05-14.htmlOpenAI shipped Codex inside the ChatGPT mobile app on iOS and Android across all plans — start, steer, approve, and review agentic coding tasks from your phone while Codex runs on your laptop, a Mac mini, or a managed remote env, with a secure relay syncing screenshots, terminal output, diffs, and…Thu, 14 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-13.htmlhttps://ericchan.io/news-feed/2026-05-13.htmlFor the first time, Anthropic has overtaken OpenAI in enterprise adoption — 34.4% vs 32.3% in April. Within 45 minutes of the data dropping, Sam Altman offered Claude Code switchers two months of free Codex; Anthropic counter-punched with a 50% raise to Claude Code weekly limits through July 13…Wed, 13 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-12.htmlhttps://ericchan.io/news-feed/2026-05-12.htmlThe FT reports Anthropic is weighing up to $50B at a $900B pre-money valuation ahead of a fall IPO — which would leapfrog OpenAI's $852B March mark. A Solana-traded pre-IPO instrument backed 1:1 by SPV exposure was implying a $1.2T valuation last week. Cerebras is upsizing its IPO with orders…Tue, 12 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-11.htmlhttps://ericchan.io/news-feed/2026-05-11.htmlOpenAI announced the OpenAI Deployment Company (DeployCo), a majority-owned subsidiary launched with $4B+ from 19 partners (TPG, McKinsey, Goldman) and an acquisition of Tomoro to staff it with ~150 forward-deployed engineers from day one. Tech Brew connects the launch to the same week's news that…Mon, 11 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-10.htmlhttps://ericchan.io/news-feed/2026-05-10.htmlSimon Willison flags a New York Times Editors' Note correcting a story that attributed a "turncoats" line to Canadian Conservative leader Pierre Poilievre — a line he never actually said. The "quote" was lifted verbatim from an AI summary that paraphrased him, and a reporter shipped it without…Sun, 10 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-09.htmlhttps://ericchan.io/news-feed/2026-05-09.htmlNate B Jones spends 28 minutes drawing the cleanest map I've seen of what actually sits between an LLM and useful work in 2026. His thesis: most people over-index on prompting, and the way to get 10× more done is to understand the harness layer — prompts, skills, plugins, MCPs, hooks, scripts — as…Sat, 09 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-08.htmlhttps://ericchan.io/news-feed/2026-05-08.htmlMozilla's collaboration with Anthropic's Mythos preview found 271 vulnerabilities in Firefox 150, up from just 22 with Opus 4.6 against Firefox 148 — a 12× jump in a single model generation. Nate B Jones argues the implication is deeper than a security headline: "a good human engineer wrote this"…Fri, 08 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-07.htmlhttps://ericchan.io/news-feed/2026-05-07.htmlAnthropic announced at "Code with Claude" that it has secured all available capacity at xAI's Colossus 1 supercomputer in Memphis — over 300 megawatts and 220,000+ Nvidia GPUs — to claw out of an 80×-revenue-growth compute crunch that had been throttling Claude Code all spring. Theo's read: this is…Thu, 07 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-06.htmlhttps://ericchan.io/news-feed/2026-05-06.htmlAnthropic's annual conference came with no new model — and a lot of platform. CPO Ami Vora opened with API volume up 17× year-on-year; Boris Cherny demoed Routines, async multi-session "higher-order prompts" that produce PR-ready output overnight; Managed Agents added multi-agent orchestration, an…Wed, 06 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-05.htmlhttps://ericchan.io/news-feed/2026-05-05.htmlOpenAI rolled GPT-5.5 Instant out as the default ChatGPT model and as chat-latest in the API, claiming a 52.5% reduction in hallucinations on high-stakes prompts vs GPT-5.3 Instant. The model becomes the first Instant-tier release flagged "High Capability" in both Cybersecurity and Bio/Chem under…Tue, 05 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-04.htmlhttps://ericchan.io/news-feed/2026-05-04.htmlOpenAI's amended Microsoft agreement kills cloud exclusivity, retires the AGI clause, and (according to Theo's reading of the fine print) flipped reverse revenue-share into a profit share — meaning unprofitable OpenAI now likely owes Microsoft "jack f***ing s***." The same day, OpenAI announced a…Mon, 04 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-03.htmlhttps://ericchan.io/news-feed/2026-05-03.htmlStripe announced hundreds of new agent commerce products that collectively represent the biggest shift in internet commerce in two decades: power is moving from sellers to buyers via AI agents. Microsoft, Meta, Visa, Mastercard, PayPal, and OpenAI are all converging on the same architecture…Sun, 03 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-02.htmlhttps://ericchan.io/news-feed/2026-05-02.htmlBloomberg reported Anthropic began fundraising talks at a valuation above $900 billion, with a $50B round expected — surpassing OpenAI's $825B March valuation and some secondary trades implying $1T. Meanwhile, the White House blocked Anthropic from expanding access to its "Mythos" model — the first…Sat, 02 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-05-01.htmlhttps://ericchan.io/news-feed/2026-05-01.htmlDeepSeek V4 Pro is a 1.6 trillion parameter model with a 1 million token context window, built by a team roughly 40x smaller than OpenAI and without top-tier NVIDIA GPUs — yet matching frontier closed models on benchmarks including a perfect 120/120 on Putnam 2025. The model and full technical…Fri, 01 May 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-30.htmlhttps://ericchan.io/news-feed/2026-04-30.htmlGitHub's reliability has cratered to 86.75% real uptime according to independent trackers, with daily outages, ~2,800 merged PRs silently reverted in a single incident, and an RCE vulnerability that could access millions of repos. Mitchell Hashimoto — GitHub user #1299, creator of Vagrant and…Thu, 30 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-29.htmlhttps://ericchan.io/news-feed/2026-04-29.htmlElon Musk's civil trial against OpenAI began in Oakland, seeking $130B in damages plus removal of Altman and Brockman. The suit — whittled from 26 claims to 2 (breach of charitable trust, unjust enrichment) — will expose private founder communications over four weeks. Meanwhile, the WSJ reported…Wed, 29 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-28.htmlhttps://ericchan.io/news-feed/2026-04-28.htmlTech Brew lays out an equation that doesn't balance: OpenAI missed its 1B ChatGPT users target for end-of-2025, missed key revenue targets, and CFO Sarah Friar is warning employees that the recent $122B raise could be depleted in three years against Altman's commitment to spend ~$600B on compute by…Tue, 28 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-27.htmlhttps://ericchan.io/news-feed/2026-04-27.htmlOpenAI's April 22 Workspace Agents launch lets any business team describe a recurring workflow in plain English and get a Codex-powered agent that runs across Slack, Drive, Calendar, SharePoint and custom MCP servers — published, scheduled, and embedded inside Slack itself. Nate B Jones argues the…Mon, 27 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-26.htmlhttps://ericchan.io/news-feed/2026-04-26.htmlGitHub Next's Maggie Appleton skewers the dominant "one man, two dozen Claudes" fantasy and argues that as implementation gets cheap, alignment — not coding speed — is the new bottleneck. She demos ACE, a research prototype that puts multiplayer chat, shared microVM cloud computers, and…Sun, 26 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-25.htmlhttps://ericchan.io/news-feed/2026-04-25.htmlTwo days after release, GPT-5.5 is sitting at #1 on Terminal-Bench (~12 percentage points ahead of Claude Opus 4.7), Artificial Analysis, LiveBench, and ARC-AGI2 (85%) — at roughly 2x the price of GPT-5.4 Extra High and a 922K-token context. OpenAI's Romain Huet confirmed there will be no separate…Sat, 25 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-24.htmlhttps://ericchan.io/news-feed/2026-04-24.htmlOpenAI's GPT-5.5 hit the top of Artificial Analysis's Intelligence Index a day after launch, but Friday's verdicts from independent reviewers were ruthless. Theo (t3.gg) called it "lazy," "card-slop"-prone, and price-gouging at $5/$30 per 1M tokens (~20% over Opus 4.7). AICodeKing's KingBench 2.0…Fri, 24 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-23.htmlhttps://ericchan.io/news-feed/2026-04-23.htmlOpenAI shipped GPT-5.5 on April 23, leading with the Codex product rather than ChatGPT, and Artificial Analysis immediately scored it the new #1 on the Intelligence Index by 3 points — first place on five of eight headline evals. Pricing doubled to $5/$30 per 1M input/output tokens (with a Pro tier…Thu, 23 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-22.htmlhttps://ericchan.io/news-feed/2026-04-22.htmlAnthropic briefly removed Claude Code from the $20/mo Pro plan, restricting it to the $100–$200 Max tiers, then rolled back after backlash — though Anthropic's Amol Avasare confirmed an A/B test hitting ~2% of new prosumer signups. Same day, GitHub paused new Copilot Individual signups, tightened…Wed, 22 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-21.htmlhttps://ericchan.io/news-feed/2026-04-21.htmlMoonshot's Kimi K2.6 is now the #1 open-weights model — a 1T-total / 32B-active MoE with a 256k window that ranks #4 overall on the AA Intelligence Index at 54, just behind the three closed frontier labs at 57. Agentic Elo jumped from K2.5's 1309 to 1520, τ²-Bench Telecom hit 96%, and the…Tue, 21 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-20.htmlhttps://ericchan.io/news-feed/2026-04-20.htmlTheo walked back a year of pushback and conceded the Opus regression is real, pointing to a BridgeMind benchmark where Opus dropped from 87.6% to 73.3% between launch and April 12. On the same day Simon Willison's updated token-counter measured the Opus 4.7 tokenizer at 1.46× more tokens on text…Mon, 20 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-19.htmlhttps://ericchan.io/news-feed/2026-04-19.htmlSimon Willison rounds up an emerging consensus — articulated by Matt Webb, Marc Benioff, and Brandur Leach — that "headless" software (API-first, no GUI) is becoming the preferred architecture for AI agents acting on behalf of users. Nate B Jones sharpens the same idea into an "agent fork" thesis…Sun, 19 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-18.htmlhttps://ericchan.io/news-feed/2026-04-18.htmlSimon Willison pulled a double shift on Anthropic's published system prompts. First, he diffed the prompt between Opus 4.6 and 4.7 — the "developer platform" got rebranded as "Claude Platform," a PowerPoint slides agent joined Chrome and Excel, and several child-safety and verbosity instructions…Sat, 18 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-17.htmlhttps://ericchan.io/news-feed/2026-04-17.htmlClaude Opus 4.7 ties GPT-5.4 and Gemini 3.1 Pro at the top of Artificial Analysis's Intelligence Index (57.0), runs away with agentic work on GDPval-AA (1,753 Elo, 79 points clear of the field), cuts hallucinations from 61% to 36%, and costs 11% less to eval than 4.6. Theo spent a full day with it…Fri, 17 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-15_2026-04-16.htmlhttps://ericchan.io/news-feed/2026-04-15_2026-04-16.htmlAnthropic released Claude Opus 4.7 on April 16 at the same price as 4.6 ($5/$25 per M tokens), posting a 13% jump on code resolution, 21% fewer errors on OfficeQA Pro, and 98.5% on XBOW's visual-acuity benchmark (up from 54.5%). Vision input resolution is tripled to ~3.75 megapixels and a new xhigh…Thu, 16 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-16.htmlhttps://ericchan.io/news-feed/2026-04-16.htmlAnthropic released claude-opus-4-7 at the same $5/M input, $25/M output pricing as 4.6, but with a new tokenizer that encodes the same prompt into up to 1.35x more tokens . Vision leaps: it accepts images up to 2,576px on the long edge (~3.75 megapixels, 3x prior Claude), scores 98.5% on XBOW…Thu, 16 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-14.htmlhttps://ericchan.io/news-feed/2026-04-14.htmlOne week after Anthropic opened Mythos to ~40 orgs via Project Glasswing, OpenAI fired back with GPT-5.4-Cyber and a scaled-up Trusted Access for Cyber program pitched to “thousands of verified individual defenders.” Simon Willison reads the “democratize access” framing skeptically — the real flow…Tue, 14 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-13.htmlhttps://ericchan.io/news-feed/2026-04-13.htmlNate B Jones argues AI-generated "dark code" — production code that no human ever understood — is not a security or tooling problem but an organizational capability problem, and it will 10x again next year. The same day, Simon Willison quotes Bryan Cantrill making a structurally identical argument…Mon, 13 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-12.htmlhttps://ericchan.io/news-feed/2026-04-12.htmlTucked into a Google/Broadcom partnership announcement: Anthropic has reached $30B in annualized revenue, a 3x jump since year-end and up 58% since February — enough to flip OpenAI on ARR for the first time. Meanwhile Meta engineers are running an internal leaderboard called "Claudonomics" to see…Sun, 12 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-11.htmlhttps://ericchan.io/news-feed/2026-04-11.htmlMorning Brew's April 11 lead is squarely Wall Street anxiety about Anthropic's newest model and its implications for banking employment — the same labor-replacement narrative that's been building all quarter, now landing on the front page of a general-business newsletter. Morning Brew blocks…Sat, 11 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-10.htmlhttps://ericchan.io/news-feed/2026-04-10.htmlTheo builds a 40-minute case, leaning heavily on Thomas Ptacek's new essay, that frontier models have quietly destroyed the economics of exploit development. The load-bearing quote: we've been safe not because software is secure, but because there weren't enough elite hackers — and now attention is…Fri, 10 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-09.htmlhttps://ericchan.io/news-feed/2026-04-09.htmlAnthropic announced Claude Mythos — a model it says is too dangerous to release generally — and launched Project Glasswing, giving exclusive access to Apple, Google, Microsoft, Cisco, and Broadcom for defensive cybersecurity work. Mythos reportedly found a 27-year-old OpenBSD bug, converted Firefox…Thu, 09 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-08.htmlhttps://ericchan.io/news-feed/2026-04-08.htmlAnthropic announced Claude Mythos Preview — a frontier model so capable at code (and, as an emergent byproduct, at finding software vulnerabilities) that they are explicitly not making it generally available. 78% on SWE-Bench Pro (up from Opus 4.6’s 53), 82% on Terminal Bench, and a 72.4% success…Wed, 08 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-07.htmlhttps://ericchan.io/news-feed/2026-04-07.htmlRyan Lopopolo, who leads frontier product exploration inside OpenAI Frontier, spent five months running a three-person team that shipped a 1M-line Electron app entirely via Codex — with a self-imposed rule that no human was allowed to write code. By January they hit 5–10 PRs per engineer per day…Tue, 07 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-06.htmlhttps://ericchan.io/news-feed/2026-04-06.htmlTech Brew surfaces a New Yorker investigation built on internal memos from Ilya Sutskever and Dario Amodei detailing how OpenAI systematically dismantled its safety commitments during the for-profit transition. The promised superalignment team got 1–2% of compute instead of the pledged 20% before…Mon, 06 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-05.htmlhttps://ericchan.io/news-feed/2026-04-05.htmlGoogle released Gemma 4, a four-size open-weights family (E2B, E4B, 24B MoE, 31B dense) under Apache 2.0, built from the same research as Gemini 3 but shrunk to run on phones, Raspberry Pis, and edge devices. It's natively multimodal (text, image, audio), ships with 128K context on the small models…Sun, 05 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-04.htmlhttps://ericchan.io/news-feed/2026-04-04.htmlTheo delivers a 40-minute polemic arguing that AI has broken the implicit contract of closed-source software: when you make every dev 100x faster, they can ship slop 100x faster too. Cursor Glass "collapsed under the weight of even the most basic changes," the Codex app ships updates that are a…Sat, 04 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-03.htmlhttps://ericchan.io/news-feed/2026-04-03.htmlSwyx and Alessio get Marc Andreessen and Jason Gson at A16Z for a wide-ranging interview in the original A16Z office, days before they move across the road. Andreessen's central claim: Pi + OpenClaw is "one of the 10 most important software" breakthroughs ever, because it marries the LLM to the…Fri, 03 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-02.htmlhttps://ericchan.io/news-feed/2026-04-02.htmlGoogle shipped four Gemma 4 models — a 31B dense, a 26B/3.8B-active MoE, and two edge models (E2B, E4B) — under an actual Apache 2.0 license, no custom clauses. The architecture is ported down from Gemini 3 research: native vision + audio, built-in thinking/reasoning, function calling, and context…Thu, 02 Apr 2026 12:00:00 GMThttps://ericchan.io/news-feed/2026-04-01.htmlhttps://ericchan.io/news-feed/2026-04-01.htmlAt 4 AM on March 31, Anthropic's claude-code npm package v2.1.88 shipped with a 57 MB source map file — over 500,000 lines of readable TypeScript that stayed mirrored across GitHub for hours before DMCA takedowns landed. Two fork projects — Python rewrite “Claw Code” and model-agnostic “OpenClaw” …Wed, 01 Apr 2026 12:00:00 GMT