The Morning Build for October 27, 2026: OpenAI's always-on 'dots', GPT-6.1 Terra price cut, and Anthropic's S-1

Today’s stories center on how AI vendors are shifting product and financial models: OpenAI turned its coding tools into an always-on assistant called dots and priced it for Pro subscribers; OpenAI also cut API prices with GPT-6.1 Terra while pushing mid-tier capabilities; Anthropic filed an S-1 that lays out record revenue, massive compute commitments, and explicit recursive self-improvement risk language; and Tesla is pitching autonomy subscriptions as a new profit engine.

OpenAI launches dots: an always-on assistant in ChatGPT for Pro and above

What happened: OpenAI released dots, an always-on personal assistant available in ChatGPT desktop and mobile to subscribers on the Pro tier or higher, with plans starting at $100 per month. Dots connects to calendar, email, computer and other tools, runs proactive research on a cadence, and inherits ChatGPT privacy controls including a user option to turn off model training.
Why it matters: Engineers integrating agents should expect persistent background activity that increases token usage and requires broader resource access and privacy controls; dots does not accept read-only cloud-drive access and currently has coarse-grained permission controls, per OpenAI’s product team comments.
Outlook: OpenAI’s ongoing conversations with payment processors, which the company says are underway, will determine whether dots gains the ability to execute payments on users’ behalf.

Sources: techcrunch.com

GPT-6.1 Terra halves OpenAI’s API prices and targets office tasks and automated software control

What happened: OpenAI released GPT-6.1 Terra at roughly half the token price of GPT-5.6 Terra, with input tokens at $1 per million and output tokens at $5 per million, and cached inputs at $0.05 per million. OpenAI positions Terra for drafting, visual reasoning, form filling and automated software control and made it available in ChatGPT Work, Codex, and via the API for Plus, Pro, Enterprise, and Edu users.
Why it matters: The model claims materially better price-to-performance for mid-range tasks and agentic automation: published comparisons cite higher Intelligence Index points per dollar, faster computer-control throughput on Agents’ Last Exam, and substantially improved coding results on Terminal-Bench 4.0 versus GPT-5.6 Terra.
Outlook: Third-party and vendor benchmark results named in coverage, Artificial Analysis’s cost-efficiency metrics, Agents’ Last Exam, Terminal-Bench 4.0, and MMMU-Pro, are the next public checkpoints for Terra’s claimed price-to-performance and hallucination reductions.

Sources: the-decoder.com

Anthropic’s S-1 frames recursive self-improvement as both a risk and a strategic advantage

What happened: Anthropic’s IPO prospectus emphasizes existential and catastrophic risks tied to recursive self-improvement (RSI) and repeatedly links RSI to faster model development, while also describing ‘closing the loop’, AI-accelerated R&D, as a core structural advantage the company intends to use to compete.
Why it matters: The filing puts public detail around how a leading frontier lab views the trade-off between accelerating model development via AI and the safeguards needed to manage those pathways; engineers and risk teams will find Anthropic’s explicit linkage of RSI to both product strategy and risk disclosures notable.
Outlook: Congressional action is the next public policy milestone referenced in the filing, including bills in Congress and at least one cosponsored by Bernie Sanders that propose limits on RSI-related practices.

Sources: wired.com · techcrunch.com · wired.com

Anthropic’s S-1 shows $11.6B Q2 revenue, high net retention, and $780B-plus compute commitments

What happened: Anthropic filed its S-1 to list on Nasdaq under ticker ANT, reporting $11.6 billion revenue in Q2 2026, a first-quarter-to-Q2 jump from $376 million in Q1 2025, net revenue retention of roughly 900 percent, and long-term compute capacity contracts totaling about $780 to $810 billion.
Why it matters: The numbers reveal a usage-driven revenue model: 80 percent billed per token, with extreme customer spending concentration metrics and fixed, large compute obligations that create a structural need for sustained high growth and justify Anthropic’s investments in Anthropic Direct data centers.
Outlook: The IPO process itself is the immediate next public milestone: pricing, offering size and the Nasdaq listing under ticker ANT led by Goldman Sachs and Morgan Stanley will be the key events to watch following the S-1 filing.

Sources: arstechnica.com

Tesla pivots to autonomy subscriptions as a path to recurring revenue

What happened: Tesla is positioning AI and software, notably its Full Self-Driving software, as a high-margin recurring revenue stream while vehicle sales continue to provide most cash flow; Elon Musk also disclosed companies are exploring a merger to accelerate the autonomy push.
Why it matters: Engineers should expect Tesla to prioritize software feature development and monetization for Full Self-Driving, while the company balances heavy AI spending against a vehicle business that still generates about 75 percent of total revenues.
Outlook: Merger exploration disclosures are the near-term corporate signal to watch, specifically any merger filings or regulatory disclosures from Tesla or the other companies Musk said are involved.

Sources: the-decoder.com · the-decoder.com