Claude Opus 4.8: More Honest, and Mythos Is Weeks Away.
Anthropic released Claude Opus 4.8 today โ better coding, parallel subagents, a cheaper fast mode, and a model trained to flag its own uncertainty. Mythos is finally weeks out.
Anthropic released Claude Opus 4.8 today โ better coding, parallel subagents, a cheaper fast mode, and a model trained to flag its own uncertainty. Mythos is finally weeks out.
Anthropic launched the Claude Compliance API on May 21 with 28 partner integrations spanning CrowdStrike, Palo Alto Networks, Okta, Zscaler, Microsoft Purview, Wiz, and IBM Guardium. It does the unglamorous job of routing Claude traffic into the same dashboards your security team already runs for everything else.
Mozilla published the first concrete result from Anthropic's Project Glasswing: 271 vulnerabilities fixed in Firefox 150, including a 20-year-old bug. Healthcare IT runs on patch cadences that are about to look very slow.
Anthropic took 300 megawatts and 220,000 GPUs at SpaceX's Memphis data center on the same day Musk testified against Sam Altman. Compute, not ideology, is now setting deal terms at the top of the AI stack.
On May 1, the Department of War expanded its classified-network AI program to eight vendors. Anthropic was not one of them and has been in federal court since March on First Amendment grounds. The procurement, the lawsuit, and the vendor silence are the same story.
Anthropic, Blackstone, Hellman & Friedman, and Goldman Sachs announced a $1.5 billion enterprise AI services firm on May 4. The dollar figure has the headlines; the target customer list โ and a near-identical OpenAI move the same day โ is the more interesting story.
Anthropic is weighing offers for a fresh round at a $900 billion valuation โ more than double February's mark and ahead of OpenAI's most recent post-money. The trajectory is real, the differentiator has a name, and the bill is for compute.
Google is putting $10 billion into Anthropic now and another $30 billion if performance targets land, at a $350 billion valuation unchanged since February. The compute loop is the point.
Axios reports the NSA is using Anthropic's unreleased Mythos model โ the one Anthropic says can find zero-days autonomously โ despite the Pentagon formally labeling Anthropic a supply-chain risk. Here's what Mythos is, who else has access, and why the operational need is winning.
Anthropic released Claude Opus 4.7 today โ better coding, sharper vision, a model that checks its own work. Its best model still isn't shipping.