OpenAI Says AI Has Overturned a 25-Year-Old Mathematical Hypothesis · 1027-772

2026-10-27 08:00

Author: RadarAI Editorial Editor: RadarAI Editorial Last updated: 2026-10-27 Review status: Editorial review pending Brief 速报官方 AI动态开源

Editorial standards and source policy: Editorial standards, Team. Content links to primary sources; see Methodology.

## 🔍 Key Insights **OpenAI** had a "scientific explosion" this week, claiming its AI made three major discoveries in **mathematics, computer science, and physics**, including refuting the roughly 25-year-old **Exponential Time Hypothesis (ETH)** and proving the existence of a quantum field theory involving QCD [0][3][16]. Meanwhile, **Anthropic**'s IPO filing revealed striking financial data: first-half revenue of **$16.3B** (a 14x year-over-year increase), but a net loss of **$80.7B**, mainly due to changes in the book value of Amazon and Google fixed-price share purchase options [4]. ## 🚀 Top Stories - **OpenAI claims AI made three cross-disciplinary scientific discoveries** [0][3][16]: covering a proof of a local smoothing conjecture, a refutation of the ETH hypothesis, and a proof of the existence of a quantum field theory, currently being verified with external experts. - **OpenAI releases GPT-6.1 Terra** [1]: focused on low-cost computer operation and visual reasoning, with a per-task cost about one-ninth that of Claude Opus 5.5. - **Anthropic S-1 filing discloses massive revenue and losses** [4]: first-half revenue of $16.3B, with the $80.7B net loss mainly from the book-value appreciation of Amazon and Google share purchase options. - **Anthropic warns of catastrophic AI risks and energy dependence** [2][17]: the IPO filing acknowledges internal concerns about the existential risks of advanced AI, and that data centers are increasingly reliant on natural gas power generation. - **DeepMind agent completes Minecraft without human intervention** [18]: 4 Gemini 4 Argon agents collaborated via chat and defeated the Ender Dragon without pre-written tools. - **DeepSeek V4.1 Flash launches on Cursor** [11]: every effort tier sits on the CursorBench cost frontier, with per-task costs from $0.60 to $0.98. - **Cohere publishes traces of Qwen3.8 anomaly investigation** [7]: releases raw tokens and minimal reproduction scripts in response to calls for transparency. - **Musk predicts robot memory demand will exceed the global total** [13]: humanoid robots buy memory per unit by body, and Tesla factories could consume the equivalent of millions of top-tier GPUs per year. ## 🔗 Sources [0] OpenAI announces three major discoveries in mathematics, computer science, and physics, and is collaborating with external experts for verification — https://aihot.news/items/g635iqfo6m6mw2m4i2e2xn1xs [1] OpenAI releases GPT-6.1 Terra: focused on low-cost computer operation tasks, with a per-task cost about one-ninth that of Claude Opus 5.5 — https://aihot.news/items/h6rsobbmnxjkeqytqpdskjxu5 [2] Anthropic discloses in its IPO filing that the company internally also has concerns about the catastrophic risks of advanced AI — https://aihot.news/items/jb1oibqp1rw1ltmm9k7cg7b7q [3] OpenAI claims AI proved results covering half of the Millennium Prize Problems, the Kakeya conjecture, and a counterexample to the Exponential Time Hypothesis — https://aihot.news/items/hhsbrhgdbx8a6v76ai2g1sgz9 [4] Anthropic S-1 filing shows first-half revenue of $16.3B and a net loss of $80.7B, mainly from fixed-price share purchase options — https://aihot.news/items/qkdi24ffzg442oh8tv0bc9zsw [7] Cohere publishes traces and reproduction scripts from the Qwen3.8 anomaly investigation — https://aihot.news

OpenAI had a "scientific explosion" this week, claiming its AI made three major discoveries in mathematics, computer science, and physics, including refuting the roughly 25-year-old Exponential Time Hypothesis (ETH) and proving the existence of a quantum field theory involving QCD [0][3][16]. Meanwhile, Anthropic's IPO filing revealed striking financial data: first-half revenue of $16.3B (a 14x year-over-year increase), but a net loss of $80.7B, mainly due to changes in the book value of Amazon and Google fixed-price share purchase options [4].

🚀 Top Stories

OpenAI claims AI made three cross-disciplinary scientific discoveries [0][3][16]: covering a proof of a local smoothing conjecture, a refutation of the ETH hypothesis, and a proof of the existence of a quantum field theory, currently being verified with external experts.
OpenAI releases GPT-6.1 Terra [1]: focused on low-cost computer operation and visual reasoning, with a per-task cost about one-ninth that of Claude Opus 5.5.
Anthropic S-1 filing discloses massive revenue and losses [4]: first-half revenue of $16.3B, with the $80.7B net loss mainly from the book-value appreciation of Amazon and Google share purchase options.
Anthropic warns of catastrophic AI risks and energy dependence [2][17]: the IPO filing acknowledges internal concerns about the existential risks of advanced AI, and that data centers are increasingly reliant on natural gas power generation.
DeepMind agent completes Minecraft without human intervention [18]: 4 Gemini 4 Argon agents collaborated via chat and defeated the Ender Dragon without pre-written tools.
DeepSeek V4.1 Flash launches on Cursor [11]: every effort tier sits on the CursorBench cost frontier, with per-task costs from $0.60 to $0.98.
Cohere publishes traces of Qwen3.8 anomaly investigation [7]: releases raw tokens and minimal reproduction scripts in response to calls for transparency.
Musk predicts robot memory demand will exceed the global total [13]: humanoid robots buy memory per unit by body, and Tesla factories could consume the equivalent of millions of top-tier GPUs per year.

🔗 Sources

[0] OpenAI announces three major discoveries in mathematics, computer science, and physics, and is collaborating with external experts for verification — https://aihot.news/items/g635iqfo6m6mw2m4i2e2xn1xs [1] OpenAI releases GPT-6.1 Terra: focused on low-cost computer operation tasks, with a per-task cost about one-ninth that of Claude Opus 5.5 — https://aihot.news/items/h6rsobbmnxjkeqytqpdskjxu5 [2] Anthropic discloses in its IPO filing that the company internally also has concerns about the catastrophic risks of advanced AI — https://aihot.news/items/jb1oibqp1rw1ltmm9k7cg7b7q [3] OpenAI claims AI proved results covering half of the Millennium Prize Problems, the Kakeya conjecture, and a counterexample to the Exponential Time Hypothesis — https://aihot.news/items/hhsbrhgdbx8a6v76ai2g1sgz9 [4] Anthropic S-1 filing shows first-half revenue of $16.3B and a net loss of $80.7B, mainly from fixed-price share purchase options — https://aihot.news/items/qkdi24ffzg442oh8tv0bc9zsw [7] Cohere publishes traces and reproduction scripts from the Qwen3.8 anomaly investigation — https://aihot.news

← Back to Updates