Blog
Developer tooling, AI model deep-dives, and a weekly roundup of the news that matters to builders.
- weeklyaillm
Dev & AI weekly — September 26, 2026
Four frontier models in 48 hours and everyone cut prices — one week after the labs said to slow down. Plus an F5 zero-day already being exploited at the network edge.
- weeklyaillm
Dev & AI weekly — September 19, 2026
The week the labs stopped shipping and started talking about not shipping: Amodei's 'pace the frontier' essay, a three-lab safety pact, and a bill with real audits.
- weeklyaillm
Dev & AI weekly — September 12, 2026
On one Thursday: DeepSeek shipped a cheaper model that beats its own Pro tier, Cursor launched agents that work while you sleep, and OpenAI put the Codex harness behind an API.
- weeklyaillm
Dev & AI weekly — September 5, 2026
Anthropic cut cache reads 75%, OpenAI shipped GPT-6 Astra, and then ChatGPT, Claude and Grok all went dark at once because they share an Azure region.
- weeklyaillm
Dev & AI weekly — August 29, 2026
Qwen open-sourced a Qwen4 architecture preview under Apache 2.0, GLM-5.3's promised weights slipped while a surprise MIT model landed, and Cloudflare's bad month got worse.
- weeklyaillm
Dev & AI weekly — August 22, 2026
The ChainDrop worm hit 400+ npm packages with 2B monthly installs, an open model took the top cyber benchmark, and Cloudflare logged 13 outages in 8 days.
- weeklyaillm
Dev & AI weekly — August 15, 2026
Gemini 3.7 Flash undercut everyone on price, OpenAI shipped a cyber model tuned to refuse less and it found real Chrome CVEs, and Meta put a 30B agent on your laptop.
- weeklyaillm
Dev & AI weekly — August 8, 2026
The keyv/cacheable npm hijack put credential-stealers in 2B downloads/month, Google put an agent inside GitHub Actions, and two more 'open' models shipped with strings attached.
- weeklyaillm
Dev & AI weekly — August 1, 2026
Kimi K3's open weights actually landed, Anthropic drew a line on open-weight policy the same week, the Claude Code team deleted 80% of its system prompt, and GitHub Actions fell over.
- product-huntlaunchdev-tools
What Product Hunt is actually for (and what it isn't)
A builder's honest read on why makers launch on Product Hunt, what you actually get out of it, and where it's overrated — with our own dev101.io launch as the worked example.
- weeklyaillm
Dev & AI weekly — July 25, 2026
Claude Opus 5 puts a cost dial on frontier reasoning, Kimi K3's open weights land at 1.4TB, and Meta's global outage revives the DNS single-point-of-failure question.
- weeklyaillm
Dev & AI weekly — July 18, 2026
Open weights went frontier-grade in one week — Inkling and Kimi K3 both landed — while the same GitHub Actions footgun that hit TanStack popped AsyncAPI.
- weeklyaillm
Dev & AI weekly — July 11, 2026
Grok 4.5 goes coding-first and cheap, npm v12 blocks install scripts by default after a year of supply-chain attacks, and Hugging Face wires one-click into SageMaker.
- weeklyaillm
Dev & AI weekly — July 4, 2026
Anthropic ships Claude Sonnet 5 for cheap agents and redeploys Fable 5 after its 19-day ban — plus critical, mostly-unpatched RCEs across AI agent frameworks.
- anthropicclaudefable-5
Claude Fable 5: How to Use It, How It Differs From Opus, and the 19 Days It Vanished
A developer's guide to Anthropic's Mythos-class Claude Fable 5 — the model ID, pricing, the forced-thinking gotcha, how it stacks up against Opus 4.8, and the export-control shutdown that took it offline for 19 days.