Tags
“Fully model-generated, without human review or triage,” is how Anthropic describes its OSS Scanner reports, and curl has rejected 19 of the 34 it has published so far. How many unreviewed AI bug reports a week could your team triage?
OpenAI’s Q3 revenue rose roughly 50% to $10 billion, per the WSJ, and operating losses stayed heavy. Anthropic made $17 billion and lost $567 million on operations. Which of the two gets your money each month, OpenAI or Anthropic?
Claude can sit in on Google Meet calls as a visible participant in Anthropic’s new beta, drafting the plan while people talk, and it leaves on its own the moment an external guest joins. On your team, who writes the meeting notes, a person or an AI tool?
November 19, 2026 is when Claude Data switches on by default for Enterprise workspaces, while the new Whiteboards and Sheets betas stay off until an admin turns them on. Would your admin let Claude Data switch on by itself or turn it off first?
TSMC’s board signed off on up to $500 million for Anthropic stock at its IPO price, and set the same cap for OpenAI preferred shares. If you invested in an AI lab, would it be public shares or a private stake?
$70-78 a share is what Anthropic expects for 825,000,000 shares as its IPO roadshow opens, though nothing can be sold until the SEC registration takes effect. Would you buy Anthropic stock at listing or wait for it to trade first?
Per Reuters, Anthropic emailed bankers IPO terms of around 867 million shares at $70-80, then wrote back that the pricing was inaccurate and the filing would be updated. When IPO numbers leak early, which do you trust more, the leak or the final filing?
About 97% of tasks passed for about $0.17 each when Opus 5.5 ran at low effort and re-ran only the failures at high. Running every task at high got 95.3% for $0.29, per Anthropic’s guide. What effort level do you run Claude at most: low, medium or high?
Pro subscribers testing Claude Fable 5.2 this week can spend up to 50% of their plan limits on it until November 16. After that, the model runs on usage credits. Is Fable 5.2 your default model this week, or just a side test?
“Forge ahead” is the WSJ’s phrase for Anthropic’s IPO plan despite AI market jitters, and people familiar say shares aim to trade Thursday or Friday next week. If Anthropic lists next week, would you buy on day one or wait a few weeks?
In an Anthropic toy run, one lead agent split Fermat’s Last Theorem among 999 helpers through a shared database, and they rebuilt the proof in 28.7 hours, against 11 days the first time. When you run parallel agents, do they share a task list or each get one job?
“Our strongest daily driver” is how Anthropic still describes Opus 5.5, even as Fable 5.2 drops to $8/$40 per million tokens, which is twice the Opus rate. For your hardest coding jobs, would you pay Fable rates or stay on Opus?
Claude Fable 5.2 arrived on Vercel’s AI Gateway with Zero Data Retention unsupported, so prompts stay stored for 30 days and teams with ZDR on get an error instead. Would you accept 30-day retention for Fable 5.2 or stay on a ZDR model?
