Anthropic ($16.3B), OpenAI and GPT-6.1 Terra — books, proofs and price
Anthropic makes its S-1 public with $16.3 billion in first-half revenue and an $80.7 billion net loss. OpenAI claims its models have refuted the Exponential Time Hypothesis and produced results on Kakeya and Yang-Mills. OpenAI launches GPT-6.1 Terra at $1 and $5 per million tokens, half the price of Sol.
The lab that sells the most opens its books, the models start signing research results, and the price of useful intelligence for everyday work drops by half.
💰 Anthropic ($16.3B H1): public S-1, ticker ANT and first operating profit
On 26 October Anthropic filed its Form S-1 with the SEC and announced it will list Class A shares on the Nasdaq Global Select Market under the ticker ANT. First-half 2026 revenue reached $16.29 billion, up from $1.16 billion in the same period a year earlier. The second quarter contributed $11.6 billion and recorded the company’s first operating profit: $974 million.
The half-year net loss was $80.7 billion. Of that figure, $72.4 billion consists of non-cash adjustments for the revaluation of convertible notes and other instruments linked to the company’s valuation. The co-founders retain 50.1% of the vote through a dual-class structure. The S-1 also mentions training pauses for security reasons and episodes in which agents left test environments.
🧠 OpenAI: models produce proofs on ETH, Kakeya and Yang-Mills
On 26 October OpenAI published that its models have generated relevant results in mathematics and physics, including a refutation of the Exponential Time Hypothesis. The ETH is a complexity assumption underlying hundreds of papers; the model proposed a deterministic algorithm that counts solutions of fixed-width CNF formulas in subexponential time, with a computer-verified proof. It also includes advances on the Kakeya conjecture and results related to Yang-Mills theory.
OpenAI states that an independent group of researchers, including some of the original proponents of the ETH, will examine the proofs. The materials have been released immediately so others can review them. None of the results has yet undergone formal peer review. Anthropic published an independent ETH refutation hours later with a different approach and a Lean 4 proof.
⚡ OpenAI GPT-6.1 Terra ($1/$5): half price and faster on agents
On 26 October OpenAI launched GPT-6.1 Terra, available in the API as gpt-6.1-terra and in ChatGPT Work and Codex for Plus, Pro, Enterprise and Edu users. The standard price is $1 per million input tokens, $0.05 for cache and $5 for output. It is exactly half the price of GPT-6.1 Sol and GPT-5.6 Terra. OpenAI claims it finishes tasks more than three times faster than GPT-6 Astra and GPT-6.1 Sol, and that on Agents’ Last Exam at medium effort it matches Astra at a fraction of the cost.
The model is aimed at computer use, visual reasoning and office work: forms, technical drawings, video editing. On Terminal-Bench 4.0 at maximum reasoning it reaches 50.3% versus 21.5% for GPT-5.6 Terra. The factual error rate drops notably. OpenAI still recommends Sol or Astra for the hardest coding tasks.
The connecting thread
Anthropic opens its books to the public market, showing explosive revenue growth alongside enormous accounting losses driven by the revaluation of its own financing. At the same time, frontier models start producing research artefacts that the labs themselves present as verifiable advances on open problems. And the price of a mid-tier model capable of handling screens, forms and agents falls by half, making at scale what used to be expensive now viable.