We gave 4 models the same prompt: build a sunset ocean. Claude Fable 5.1: $1.67, 6m 20s Claude Opus 5.5: $0.90, 6m 45s GPT-6 Astra: $0.59, 3m 54s GPT-6 Sol: $0.07, 44s Which one did the best?

Today is the 10 year anniversary of Founders. It’s my life’s work. Thank you for listening.

We're publishing our most detailed threat intelligence report to date. It covers how people tried to misuse Claude—for cyberattacks, influence operations, surveillance, biology, and building weapons—and how we found and stopped them. We disrupted every operation in the report, and used the lessons from them to strengthen our safeguards. Where appropriate, we also shared what we found with authorities and other AI companies. These cases are not typical: we’re highlighting some of the most sophisticated misuse we’ve seen. But they’re especially important to discuss, because they show us where AI misuse is headed, where our safeguards work, and where they need to improve. We’re publishing this report so others can spot the same activity on their own platforms, and so we can give the public a clearer view of how emerging threats develop. Read the report: anthropic.com/threat-intelli…

Breaking my X hiatus because this is fascinating. I ran threat disruption at Meta for 8 years, where we routinely released reports like the Anthropic report out today. It’s noteworthy that Anthropic is being this transparent, and they deserve credit if we want this level of visibility into AI companies going forward. Here's what stood out to me, and where I'd push back. 🧵

Same prompt run through GPT 6 Astra in Codex at all six effort levels. Low: 7,923 tokens, 4 min, $0.63 Medium: 9,085 tokens, 5 min, $0.69 High: 19,596 tokens, 10 min, $1.21 Extra high: 34,420 tokens, 19 min, $1.95 Max: 37,241 tokens, 20 min, $2.09 Ultra: 27,465 tokens, 14 min, $1.84 Max used nearly 5x the tokens of low and produced the best result. Full GPT 6 Astra outputs from every level in the thread below.