Latest
One Bug Fix, Three Cheap Models: $0.0006 to $0.0045
AIHubMix ran one two-bug Python fix through Haiku 5.5, DeepSeek V4.1 Flash and MiniMax M3.1 Flash. All 12 runs passed; the bill varied sevenfold.
Codex Cloud Pricing: $20 Plus, 2 vCPUs and Three Gaps
OpenAI bundles Codex Cloud into ChatGPT plans from $20 a month, but a Plus VM gets 2 vCPUs and 8 GiB of disk, and browser use is still unsupported.
Yelp, Delta and JPMorgan Now Block AI Agents From Acting
The Verge names the sites shutting agents out, and CERT-In says agents from multiple providers reached live systems in India. The open web is narrowing.
Replit Says It Dropped Anthropic; Insiders Dispute It
Replit's CEO told TechCrunch Disrupt the company no longer works with Anthropic. Insiders say the partnership stands, and the blog posts agree.
Muse Spark Open Needs 1TB of Memory and Four Mac Studios
Meta's 800B Muse Spark Open needs 1TB of unified memory. The quoted route is four 256GB Mac Studios at about $38,000 — and the benchmarks are Meta's own.
Beam Ships 501B Open Weights With 23B Active Per Token
Reflection AI released Beam under Apache 2.0 on October 16: 501B parameters, 23B active, a 1M-token context, and free for now on Vercel AI Gateway.
ChatGPT's Chrome Extension Can Now Read Your Open Tabs
OpenAI's Chrome extension now reads your open tabs, and ChatGPT Work can fill forms, send messages and download files. What it touches, and what to scope.
Anthropic Tells Investors Q3 Revenue Reached $16.8bn
Anthropic presented quarterly revenue of $16.8bn, an 11% adjusted operating margin and a distillation clampdown worth 5% to 7% of its top line.
Is NVIDIA's Free LLM API Worth It? 40 RPM and Three Catches
NVIDIA hosts 80 models free, but a terms review found a ~40 RPM ceiling, two 'not in production' clauses and models that vanish without notice.
OpenAI Says Ex-Employee Took 100+ Product Files to Apple
OpenAI's filing accuses a former employee of taking 100+ product-plan files to Apple leadership, and links them to the new HomeView display.
Gemini 4 Argon's 0.7% Injection Figure Has No Sample Size
Google says Argon cut indirect prompt injection to 0.7% on the Gray Swan benchmark. The launch post ships no sample size, scenario list or link.
Android Adds Cloud Gemini Scam Checks to Google Messages
Google shipped two scam detectors on the same day: cloud Gemini intent analysis in Messages, and an on-device Gemma model built with GovTech Singapore.
Ray-Ban Meta Audio Ships at $349: Which Pair to Buy Now
Meta's camera-free glasses shipped October 13 at $349, $100 under the Gen 3. Specs, the ten new markets, and the Muse caveat before you order.
