Startups

Anthropic ships Opus 4.8, teases broad Mythos release, and closes $65B at a $965B valuation

A faster, more honest flagship model and the imminent end of Mythos's restricted preview land on the same day as a record Series H that vaults Anthropic past OpenAI.

By Giulia Donatello · Startups & funding · 2026-05-30 · 4 min read

Photo: Unsplash — Abstract visualization of an AI language model

Anthropic disclosed a $65 billion Series H at a $965 billion post-money valuation on Thursday, the same day it shipped Claude Opus 4.8 and signaled that its restricted Mythos preview will open broadly within weeks. Three announcements, one orchestrated press cycle, and a valuation that now sits comfortably ahead of OpenAI’s last reported mark.

The financial arithmetic is the part worth pausing on. In February, Anthropic closed a $30 billion Series G at $380 billion. Three months later, the post-money has more than doubled. SiliconAngle reports the company’s run rate more than tripled to $47 billion in that same window, against roughly $1 billion annualised at the end of 2024 and an estimated $30 billion run rate earlier this year. Growth at that slope is what justifies a near-trillion-dollar private mark to the funds writing the checks; whether it justifies it to anyone else is a different question.

Opus 4.8 lands 41 days after 4.7, priced at $5 per million input tokens and $25 per million output. Anthropic claims it’s roughly four times less likely than 4.7 to let flaws in code pass unremarked, with Terminal-Bench 2.1 rising from 64.3% to 69.2% and multidisciplinary reasoning with tools moving from 54.7% to 57.9%. A new Dynamic Workflows tool is pitched at orchestrating hundreds of parallel subagents, and Claude Code now claims codebase-scale migrations across hundreds of thousands of lines. The upgraded fast mode is 2.5 times quicker and three times cheaper than its predecessor.

The “most honest model yet” framing is doing real work here. Coding assistants that quietly wave through bad code are the unspoken liability of every enterprise rollout, and honesty metrics are how Anthropic is differentiating against Google, Microsoft, AWS, and Apple’s in-house efforts.

Then there’s Mythos. The preview, run as Project Glasswing with roughly 50 partners, has surfaced more than 10,000 high- or critical-severity vulnerabilities across critical software infrastructure. That number is the actual pitch. A general release lets Anthropic convert security telemetry into a procurement argument no rival can match on the same day.

Coordinated launches like this aren’t accidents. They’re the closest thing private-market AI has to an earnings narrative, and Anthropic just delivered three quarters’ worth in a single morning.

Sources