律动BlockBeats|Nov 22, 2026 04:08
[OpenAI Suffers Epic Triple Blow: GPT-6.2 Failure, Google Neon Gains Popularity, Claude Proves Riemann Hypothesis] Beating AI News reports that AI blogger Binkg posted that, according to a friend working at OpenAI, morale within the company is low, and they may be experiencing their most difficult week since the company's founding.
On November 17th, OpenAI released GPT-6.2 Astra and Sol, but it was met with sharp criticism from several early testers. Renowned developer Theo bluntly stated that while Astra is intelligent, it is difficult to use in practical development. He once ran the model continuously for 11 days, attempting to optimize a Rust compiler, but failed to submit any effective improvements and even caused the program's execution speed to drop by 56% at one point.
Theo also complained that Astra often misses the mark, gets stuck in invalid loops when encountering complex tasks, and then uses a lot of obscure jargon to explain simple problems after completing the work. OpenAI product manager Tibo responded that the early tests used an older model, and significant modifications had been made before the official release. Theo acknowledged improvements after retesting, but remained unsatisfied.
Meanwhile, Google launched Gemini 4 Neon the following day. The new model's natural conversational style and programming capabilities won praise from many developers. Binkg believes that Google had already overtaken OpenAI that week.
Anthropic stole the spotlight in mathematical research. On November 20th, researcher Levent Alpöge announced a collaboration with Claude to prove the Riemann Hypothesis, a problem that had baffled the mathematical community for over 160 years, and published approximately 17 million lines of formalized Lean proof code. The following day, the Lean and Mathlib communities jointly issued a statement confirming the proof's validity after verification by multiple independent tools.
OpenAI's financial situation is also precarious, with the company reporting an operating loss of $10.9 billion in the third quarter. Its latest funding round, originally planned to raise up to $30 billion with a target valuation of $1.5 trillion, has now been reduced to $1.25 trillion, with only $11 billion raised in the first tranche.
Binkg concluded by asserting that OpenAI no longer holds a single, dominant position in any area, and "it may really be finished." [Original Link]
Share To
Timeline
HotFlash
APP
X
Telegram
CopyLink