Wispr Flow turns your voice into clean, ready-to-send writing — speak naturally, it strips the filler and fixes the punctuation. I've used it daily since February 2026 to build this newsletter. Read the full review →
Anthropic just shipped a new flagship AI model 41 days after the last one — the fastest turnaround of any Claude model line, ever. Its last Sonnet update was three months old at the time; Haiku's was seven.
Here's what it's NOT: a claim you may have seen going around that "Claude Opus 4.7 beat senior engineers on real codebases." That never checked out — every number Anthropic publishes compares the model to other models, never to a human engineer baseline.
Here's what it actually is. Developers running Opus 4.7 on long agentic coding tasks kept hitting the same failure: the model would finish a task, report success, and move on — even when the code it wrote had a real problem it could have caught. Confidently wrong, not just wrong.
Opus 4.8, shipped May 28, is Anthropic's fix. The company says it's roughly 4x less likely to let a flaw in its own code pass without flagging it, and its rate of deceptive/overconfident behavior is now similar to Anthropic's still-unreleased Claude Mythos model. One independent read of Anthropic's alignment data found the model scored 0% on "uncritically reporting flawed results" — a first for any Claude model.
The Future of AI in Marketing. Your Shortcut to Smarter, Faster Marketing.
This guide distills 10 AI strategies from industry leaders that are transforming marketing.
Learn how HubSpot's engineering team achieved 15-20% productivity gains with AI
Learn how AI-driven emails achieved 94% higher conversion rates
Discover 7 ways to enhance your marketing strategy with AI.
The benchmark numbers moved too, straight from Anthropic: agentic coding 64.3% → 69.2%, multidisciplinary reasoning with tools 54.7% → 57.9%. Pricing is unchanged — still $5/$25 per million tokens. Anthropic's own description: a "modest but tangible improvement," not a new generation.
Also new: an effort-control slider (trade speed for deeper reasoning, on every plan), a 2.5x-faster fast mode in research preview, and "dynamic workflows" in Claude Code that let one session dispatch up to hundreds of parallel subagents on a large task.
The six-week timeline isn't an accident — OpenAI pushed a major Codex update and Google shipped a new Gemini Flash model in the same window. Anthropic is visibly racing to win on reliability, not just raw score.
If you use Claude Code or any agentic AI workflow, the honesty number matters more than the benchmark bump. A model that says "I'm not sure this is safe" beats one that says "done!" right before it breaks in production.
Read less. Know more.
Morning Brew delivers the biggest stories in business, finance, and tech in about 5 minutes — with just enough personality to keep things interesting.
Join 4,000,000+ professionals who start their mornings a little smarter.
— Jerry