What a vibe check is here A vibe check is the Every team's name for hands-on, informal testing of a newly released AI model — putting it through real writing, coding, and knowledge-work tasks to build an evidence-backed picture of what it is genuinely good at, where it falls short, and how it compares to rival models. It exists because the answer to is this model better for your work? can't be read off a spec sheet or leaderboard; it comes from actually using the model. How the team runs them Across a year of releases, the team has repeated the practice in a consistent set of forms: What a vibe check produces A vibe check ends in a verdict a reader can act on: which benchmarks and informal tests are worth running, where specific models — Claude Opus variants, GPT-5.x, Codex, Fable, Astra — excel or fall short, and how to combine models so their blind spots cover each other. Notably, when the team disagreed during a session, the published chapter preserves that disagreement rather than flattening it into a single score. • A vibe check is hands-on, informal evaluation of a new AI model on real tasks, not benchmark reading. • It takes the form of day-zero reviews, live release-day streams, and panel discussions. • It produces evidence-backed strengths, weaknesses, and comparisons — including preserved disagreements — rather than a single score.