Anthropic开源FlashProve：CPU可验证AI模型身份
Anthropic just open-sourced FlashProve, which lets you check which AI model actu…
Anthropic just open-sourced FlashProve, which lets you check which AI model actually wrote a reply
Anthropic 刚刚开源了 FlashProve，它可以让您检查哪个 AI 模型实际上撰写了回复
checkable on an ordinary CPU.
在普通 CPU 上即可验证。
e.g. if you pay for a top AI model through a reseller, you can’t check whether it quietly served a cheaper one. Anthropic’s FlashProve lets the server prove mathematically which model answered.
例如，如果您通过转售商付费使用顶级 AI 模型，则无法检查它是否悄悄提供了更便宜的模型。Anthropic 的 FlashProve 允许服务器从数学上证明是哪个模型进行了回答。
Anthropic tested it on Meta’s Llama 4 Maverick, a large 400B-parameter open model, where a single Nvidia H100 GPU produced proofs at 110 tokens/second, each about 19MB.
Anthropic 在 Meta 的 Llama 4 Maverick（一个拥有 4000 亿参数的大型开源模型）上对其进行了测试，其中单个 Nvidia H100 GPU 以每秒 110 个 token 的速度生成证明，每个证明约 19MB。
A regular CPU checks each proof in about 17 seconds, while VerInf, the only earlier system known to prove this model, made 93.6GB proofs that took almost a day to check.
普通 CPU 检查每个证明大约需要 17 秒，而此前已知的唯一能证明该模型的系统 VerInf 生成的证明高达 93.6GB，检查几乎需要一天时间。
Claude built the system, and because its code is hard to audit, the security guarantees are mathematically proved in Lean 4, a proof-checking language,
Claude 构建了该系统，由于其代码难以审计，安全保证已在 Lean 4（一种证明检查语言）中通过数学方式得到证明。
更进一步：量化金融体系
看懂新闻只是起点——沿量化金融路径，把它变成能交付的工程能力