有人算 GPT-6 Astra 解数学难题的成本
他估算 Astra 跑几小时解掉悬置十年的数学题，总成本约 800 美元，认为完全塞得进 500 美元月费。
I want to dwell on this a bit. Some of these problems were tackled by some of the smartest humans on the planet for maybe 10+ years and they couldn't solve them. Now a model worked for 3 hours per problem and solved 100s of them. Even assuming that ChatGPT Pro is ~10 sub-agents working constantly, that would perhaps be ~300m tokens total or $800 at Astra API prices. Even if this LLM would be priced at 10x, this is still hardly a huge amount of money and certainly fit within a $500/month sub. To say this again, you can wake up in the morning, pick a maths problem that hasn't been solved in decades, put it to your LLM, the make yourself a coffee, walk your dog, check your emails, smoke a joint... and the problem could be solved by the time you check back in. This is not normal.
这条是引用着另一条写的：x.com/petergostev/status/2107610428683538854