Axios reports the NSA is using Anthropic's Mythos Preview even as Pentagon officials call the company a supply-chain risk. The clash puts AI safety limits, federal cyber demand, and procurement politics in the same room.
HN upvoted this because it turned vague limit anxiety into numbers. Tokenomics says 541 anonymous submissions averaged 466 request tokens on Opus 4.7 versus 349 on Opus 4.6, a 38.1% increase, and the thread immediately argued over what that means for real Claude usage.
Why it matters: Anthropic is moving Claude into the document surface where legal, finance, and policy work already happens. The beta covers Pro, Max, Team, and Enterprise users and keeps edits as native Word tracked changes.
Why it matters: Anthropic is moving Claude into visual work products, not just text and code. The tweet says Claude Design is powered by Opus 4.7 and is rolling out in research preview to Pro, Max, Team, and Enterprise plans.
Why it matters: Anthropic is pushing Opus toward longer autonomous coding work without raising the premium model price. The linked launch page says Opus 4.7 reaches 70% on CursorBench versus 58% for Opus 4.6, while API pricing stays at $5 per million input tokens and $25 per million output tokens.
Claude Opus 4.7 is now generally available across Claude products, the API, Amazon Bedrock, Vertex AI, and Microsoft Foundry. Anthropic kept pricing at $5/$25 per million tokens while adding higher-resolution image handling, xhigh effort, and stronger coding-agent behavior.
Reuters’ new Mythos analysis argues banks are staring at a timing problem, not a distant risk. Officials in the U.S., Canada, and Britain have already met with banking leaders, and Anthropic says the model found thousands of high and critical vulnerabilities.
This is less about prestige and more about control of the board. Anthropic says Vas Narasimhan, the Novartis CEO who has overseen more than 35 novel medicines, is joining through the Long-Term Benefit Trust, pushing trust-appointed directors into the majority.
Anthropic is pushing Claude Code beyond one-off coding sessions and into persistent workflow automation. In research preview, routines can launch from 3 trigger types—schedules, API calls, and GitHub events—and are available across 4 paid plan tiers when Claude Code on the web is enabled.
Anthropic is using Claude not just as a model to align, but as a researcher that improved weak-to-strong supervision nearly to the ceiling. In the linked study, nine Claude Opus 4.6 agents pushed performance-gap recovery from a 0.23 human baseline to 0.97 after 800 cumulative research hours.
A Reddit thread pulled attention to AISI’s latest Mythos Preview evaluation, which shows a step change not just on expert CTFs but on multi-stage cyber ranges. The important claim is not generic danger rhetoric, but that Mythos became the first model to complete a 32-step corporate attack simulation end to end.
Anthropic unveiled Project Glasswing on April 7, 2026, giving major tech and security partners access to Claude Mythos Preview for defensive vulnerability discovery. The company says the model has already found thousands of high-severity flaws and is backing the effort with up to $100 million in usage credits and $4 million in open-source donations.