In partnership with
Hello from The AI Night,
Today in AI:
Anthropic Launches Claude Opus 4.7 for Developers
Perplexity Releases Personal Computer for Mac Users
OpenAI Debuts GPT-Rosalind for Drug Discovery Research
Here's the deal: Anthropic Launches Claude Opus 4.7, a direct upgrade to Opus 4.6 focused on hard software engineering, higher resolution vision and a controlled rollout of cyber capabilities ahead of its Mythos Preview model.
The Breakdown:
Handles long running coding tasks with closer instruction following and self verification before reporting back.
Accepts images up to 2,576 pixels on the long edge (~3.75 megapixels) over 3x prior Claude models.
State of the art on the Finance Agent benchmark and GDPval-AA, a third-party knowledge work eval.
Adds an xhigh effort level between high and max; Claude Code defaults to xhigh and adds /ultrareview plus auto mode for Max users.
Pricing unchanged at $5/$25 per million input/output tokens; new tokenizer may raise token counts 1.0 to 1.35x.
Automated safeguards block high-risk cyber requests; researchers can apply to the new Cyber Verification Program.
The bigger picture: Anthropic is doing something no other lab has tried publicly. They are using a production model as a live stress test for cyber safety controls before releasing something far more dangerous. If these guardrails hold under real world usage, Mythos gets a green light. If they break, everyone learns something uncomfortable about where the ceiling actually is.
Here's the deal: Perplexity Launches Personal Computer, a feature integrated with the Perplexity Mac App that orchestrates tasks across local files, native Mac applications and the browser. It was announced on 11 March 2026 and is rolling out today to all Perplexity Max subscribers and everyone on the waitlist.
The Breakdown:
Connects to any folder to search, read and write files locally.
Works across iMessage, Apple Mail, Calendar and other native Mac apps.
Runs 24/7 in the background when set up on a Mac mini, operating across apps and files.
Supports remote task initiation; start a task from iPhone and Personal Computer executes on the desktop using 2FA.
Requires the latest iOS update from the App Store and the updated Perplexity Mac app.
The bigger picture: Anthropic has Cowork, Apple has Siri with app control. Now Perplexity is fighting for the same territory with zero OS-level access. Running 24/7 on a Mac mini with remote iPhone triggers is clever but Perplexity is betting it can out-execute companies that literally own the operating system. That is a bold bet.
Here's the deal: OpenAI released GPT-Rosalind, the first model in a new life sciences series optimized for biology, drug discovery and translational medicine. It is available as a research preview in ChatGPT, Codex and the API for qualified Enterprise customers in the U.S. through a trusted access program.
The Breakdown:
Tuned for multi-step scientific workflows including literature review, sequence to function interpretation, experimental planning and data analysis.
Scored 0.751 Pass@1 on BixBench, ahead of GPT-5.4 (0.732), Grok 4.2 (0.698), and Gemini 3.1 Pro (0.550).
Outperformed GPT-5.4 on 6 of 11 LABBench2 tasks with the largest gain on CloningQA.
In a Dyno Therapeutics RNA evaluation, best of ten submissions ranked above the 95th percentile of human experts on prediction and around the 84th percentile on sequence generation.
Ships with a free Life Sciences research plugin for Codex connecting to 50+ scientific tools and databases.
Launch partners include Amgen, Moderna, Thermo Fisher, Novo Nordisk and the Allen Institute.
The bigger picture: OpenAI is no longer building one model for everyone. Rosalind is the first vertical specific frontier model from a major lab with real pharma partners behind it. If Moderna and Amgen start publishing results using this every competing lab will need its own domain model or risk falling behind in discovery speed.
AI Agents Are Reading Your Docs. Are You Ready?
Last month, 48% of visitors to documentation sites across Mintlify were AI agents—not humans.
Claude Code, Cursor, and other coding agents are becoming the actual customers reading your docs. And they read everything.
This changes what good documentation means. Humans skim and forgive gaps. Agents methodically check every endpoint, read every guide, and compare you against alternatives with zero fatigue.
Your docs aren't just helping users anymore—they're your product's first interview with the machines deciding whether to recommend you.
That means:
→ Clear schema markup so agents can parse your content
→ Real benchmarks, not marketing fluff
→ Open endpoints agents can actually test
→ Honest comparisons that emphasize strengths without hype
In the agentic world, documentation becomes 10x more important. Companies that make their products machine-understandable will win distribution through AI.
What else you need to know:
Google's Gemini app integrated Personal Intelligence with its Nano Banana 2 image model, letting the assistant factor in user preferences and interests when generating images.
Anthropic's Claude team introduced @ClaudeDevs on X, a developer-focused account sharing changelogs, API releases, community updates and technical deep dives for those building with Claude.
Vercel released Workflows to general availability, letting developers use code as the orchestrator to run agents, backends, and long running processes without managing queues, retries, or workers.
Cursor partnered with University of Chicago economist Suproteem Sarkar to study 500 teams, finding a 68% increase in high complexity development tasks as models became more capable this year.
Google upgraded AI Mode in Chrome, letting desktop users open linked pages side by side with Search and pull context from multiple tabs, images or PDFs into queries.
That’s it for today’s edition of The AI Night.
Our goal is to cut through the noise, surface what actually changed, and explain why it matters.
3 ways to support us:
Forward this to your AI-curious friend → https://www.theainight.com
Sponsor The AI Night and reach 500+ AI builders daily → passionfroot.me/theainight
Reply to this email — I read every response