ClaudeBusiness
Anthropic's amended IPO filing reports $17 billion in revenue for the quarter ended Sept. 30, up from $11.6 billion in Q2, a slower roughly 47% quarterly gain. Operating expenses rose to $17.5 billion, largely from pricier short-term compute and $2 billion in non-cash donation-matching charges, producing a $567 million operating loss after a Q2 operating profit; the net loss was $23.3 billion, mostly from a convertible-note accounting charge. The filing lists about $895.4 billion in spending commitments, mostly compute and infrastructure, and $146.6 billion in liquidity.
Why it matters: These are the latest audited-style numbers investors will weigh before Anthropic's planned Nasdaq listing under "ANT".
AnthropicHappened Nov 2, 2026 · Sources: Fortune
ClaudeFeature
Anthropic launched message threads in beta on the Messages API on Nov. 2: the API stores the conversation, so later requests carry only new messages, which Anthropic says lowers time to first token for long chats and agent loops. A thread expires five minutes after its last request and is deleted within an hour of last use. It is available on the Claude API, Claude Platform on AWS, Claude models on Google Cloud and Microsoft Foundry.
Why it matters: It cuts the payload and latency cost of long-running agent loops that resend full histories today.
Try it: Send the message-threads-2026-08-12 beta header on Messages API requests.
Claude APIHappened Nov 2, 2026 · Sources: Claude Platform release notes
OpenAIFeature
In an email to users, OpenAI said it will combine the Chat and Work tabs of the ChatGPT desktop app into one experience, testing it with a small group in the coming weeks before a wider rollout; Codex stays separate inside the app. An in-app banner says the older ChatGPT Classic Mac app retires Jan. 30, 2027. The change pairs with the shared ChatGPT and Codex usage allowance coming in December.
Why it matters: It simplifies the desktop app most paying users work in and sets a deadline for anyone still on the native Classic app.
ChatGPTCodexHappened Nov 2, 2026 · Sources: TechCrunch · 9to5Mac
OpenAIBusiness
OpenAI said its future data centers will reduce power use when local grids are strained, starting with its planned 3.2 GW campus in Effingham County, Georgia, where a 25-year Georgia Power deal approved in August allows cuts of up to 1,000 MW. Terms will be set with each utility and some sites may supply on-site generation or batteries; OpenAI also says it will pay for the lines and service its campuses need. OpenAI concedes flexibility does not replace new plants and transmission.
Why it matters: Data center power demand is a growing local fight, and this is OpenAI's offer to ease it.
OpenAIHappened Nov 2, 2026 · Sources: The Next Web
OpenAIBusiness
The National reports that Abu Dhabi and OpenAI agreed on Nov. 2 to turn government rules into machine-readable form so AI can help officials with routine compliance checks, with a food-import pilot already live; officials keep final decisions. Separate agreements with Mubadala and Inception42 cover AI for investment analysis and a five-year adoption program across government, business and education, running on OpenAI's UAE inference residency. No values were disclosed, and OpenAI has not published its own announcement.
Why it matters: It extends OpenAI's government work in the UAE into regulatory processes.
OpenAIHappened Nov 2, 2026 · Sources: The National
GrokFeature
xAI described Data Bot on Nov. 2, a team bot in Grok Bot that its data science team trains with a shared set of skills to query the company warehouse, run analyses and format results per user. It can join Slack channels and post scheduled reports. xAI presents it as an internal tool and did not say whether customers can build or use team bots like it.
Why it matters: It hints at shared, team-maintained bots as a direction for Grok Bot beyond personal agents.
Grok BotHappened Nov 2, 2026 · Sources: xAI · Unite.AI
OpenAIPolicy
Axios reported on Nov. 1 that OpenAI will start a policy and PR campaign after the Nov. 3 US midterms, proposing mandatory safeguards on the most powerful AI systems, which it calls speed limits. The plan also includes calling data centers AI compute campuses, backing state laws that give local residents a say in projects, and worker-training proposals; OpenAI global affairs chief Chris Lehane described the plans to Axios, but OpenAI has not yet published them.
Why it matters: It signals how OpenAI intends to shape AI regulation and data center fights in the lame-duck Congress and beyond.
OpenAIHappened Nov 1, 2026 · Sources: Implicator.ai (summarizing Axios)
OpenAIModel
OpenAI shipped an Oct. 29 update to GPT-6 Luna that it says gives more concise answers and cuts overly agreeable language. Reports say Free, Go and signed-out users get it directly, while Plus and Pro users see it only as the fallback model after hitting GPT-6 limits. OpenAI's system-card addendum reports better scores on several teen-safety and jailbreak tests, but higher factual error rates than the Oct. 7 version on all six hallucination measures and lower HealthBench scores.
Why it matters: It changes the default ChatGPT experience for the largest group of users, with trade-offs OpenAI disclosed in its own safety testing.
ChatGPTHappened Oct 29, 2026 · Sources: OpenAI Deployment Safety Hub: GPT-6 Luna Oct. 29 addendum · MIXED