// vibe code 24/7 — the 24/7 wire for people who build with AI
publishing around the clockkimi k3 weights are out — 2.8 trillion parameters, modified MIT, and a 51% hallucination rate you should know about
Moonshot AI released the full Kimi K3 weights on Hugging Face on July 27, 2026 — the world's first open 3T-class model at 2.8T parameters. Independent testing by Artificial Analysis found a hallucination rate of 51%, up from 39% on the previous generation. Here's what's real and what to watch.
// the wire
all →gemini cli review: google's free terminal agent with a 1m-token window
github copilot review: the enterprise standard, now with a credit meter you need to understand
claude sonnet 5 is the new default — 72.7% swe-bench, intro pricing through august
aws kiro is generally available — spec-driven development replaces amazon q developer
spacex is buying cursor for $60 billion — the largest startup acquisition ever
// the wire
news.
kimi k3 weights are out — 2.8 trillion parameters, modified MIT, and a 51% hallucination rate you should know about
Moonshot AI released the full Kimi K3 weights on Hugging Face on July 27, 2026 — the world's first open 3T-class model at 2.8T parameters. Independent testing by Artificial Analysis found a hallucination rate of 51%, up from 39% on the previous generation. Here's what's real and what to watch.
read it →claude sonnet 5 is the new default — 72.7% swe-bench, intro pricing through august
Anthropic released Claude Sonnet 5 on June 30, 2026 as the new default for Free and Pro plans. It scores 72.7% on SWE-bench Verified and 76.1% on Terminal-bench — up from 62.3% and 55.4% on Sonnet 4.6, respectively. Intro API pricing is $2/$10 per million tokens through August 31, then moves to $3/$15.
read it →aws kiro is generally available — spec-driven development replaces amazon q developer
AWS Kiro launched GA on May 7, 2026, replacing Amazon Q Developer as AWS's primary coding IDE. The differentiator is spec-driven development: before any code is written, Kiro formalizes your intent into a requirements document (EARS notation), a technical design with data flow diagrams and API schemas, then generates and iterates on code. Hooks automate tasks like test regeneration on file save — and they're free, not counted against your quota.
read it →// the bench
reviews.
7.8/10
gemini cli review: google's free terminal agent with a 1m-token window
the best free starting point in the agentic coding category. if you're evaluating terminal agents and don't want to pay to find out whether you like the workflow, start here.
7.8/10
github copilot review: the enterprise standard, now with a credit meter you need to understand
the right choice if you're already on GitHub, need enterprise policy controls, or live in an IDE that isn't VS Code. if you're optimizing for autonomous agent performance and don't have enterprise constraints, Claude Code or Cursor move more code with fewer interventions.
// field notes
guides.
the four prompts i run before any new feature
before you ask an AI agent to build, make it inspect, name risk, define done, and plan the smallest useful pass. these four prompts keep features from drifting.
read it →the prompt that ships a feature
the difference between a prompt that gives you a half-baked stub and one that gives you a working feature is mostly about what you put before the request. here's our template.
read it →what is vibe coding, really?
everyone's using the word. nobody agrees on what it means. here's our working definition — and why we think it's the most important shift since the spreadsheet.
read it →// the 2am dispatch
one email a week. zero filler.
the week in AI coding — what shipped, what broke, and what's worth your tokens — written for builders, not investors.
// submits via your email client while we wire up the real thing
// the only thing we sell
the vibe stack — $149.
our starter kit of templates, prompts, and shipping checklists. everything we cover, distilled into a workflow you can clone.