AIsaAIsaStudio English 打开 AIsa Studio

前 Anthropic 研究员 Coxon:现在只需一小部分计算量即可重新训练 Mythos

原标题:Anthropic can already retrain Mythos with "a fraction of the compute originally needed," says former researcher Coxon

前 Anthropic 与 OpenAI 研究员 Jacob Coxon 表示,Anthropic IPO 路演的一张幻灯片是他在入职约四个月后离职的关键原因。该图以对数坐标展示代码预训练计算乘数随时间的变化,他据此认为算法效率正以超指数速度加速,Anthropic 现在用远少于原来的计算量就能重新训练 Mythos 模型,且算法改进接近被自动化,因此主张明年暂停前沿模型开发,让竞争对手追赶。Anthropic 注明这些数据点基于内部估算、仅供示意。

为什么值得看

前研究员基于 Anthropic IPO 路演幻灯片给出的内部估算解读,读者可以借此了解算法效率加速及其暂停前沿训练的主张。

阅读原文 ↗

本页的摘要、推荐理由与解读出自 AIsa Studio 的 AI 风向智能体,原文版权归原作者所有。

原文

Jacob Coxon, former OpenAI and Anthropic researcher who brought AI fears into the mainstream, says a slide from Anthropic's IPO roadshow was a major reason he quit.

Coxon says a slide Anthropic released for its IPO roadshow was a key reason he left the company after only about four months. The chart shows how much compute Anthropic needs over time to train its models, and the axis is compressed so that a 10x and a 100x increase barely look different.

Yet the curve still climbs steeply, which Coxon says shows that algorithmic efficiency is improving not just quickly but at an accelerating rate. "It's an insane plot and I'm surprised they made it public," he writes.

Coxon argues that Anthropic's Mythos model, a major capability jump in the spring, was part of this accelerating trend. "Mythos was not a one-time deviation, but a point on a superexponential. They could already retrain Mythos with a fraction of the compute originally needed," he writes.

This slide from Anthropic's IPO roadshow shows the "Pre-training compute multiplier" for code over time. The y-axis is logarithmic. Former Anthropic researcher Jacob Coxon says the steep climb from July 2024 to October 2026 points to superexponential algorithmic progress. Anthropic notes that the data points are based on internal estimates and are for illustration only. | Image: Anthropic

He says algorithmic improvements are "on the verge of being automated," which is why he wants frontier model development paused next year so competitors can catch up. A pause, he writes, would "commoditize human-level intelligence. It would only harm OpenAI and Anthropic."

Coxon previously trained AI models at OpenAI and, most recently, Anthropic. He drew wide attention in September with public warnings about the risks of advanced AI systems, echoed by other protagonists in the industry. One of their fears is that advanced, self-improving AI could get out of control.

AI News Without the Hype – Curated by Humans

Subscribe to THE DECODER for ad-free reading, a weekly AI newsletter, our exclusive "AI Radar" frontier report six times a year, full archive access, and access to our comment section.

Subscribe now

原文版权归 The Decoder 及原作者所有,本站转载仅供阅读。如需删除,请联系 support@aisa.one。

事件综述

Anthropic 启动 IPO:发行价 70–78 美元,目标估值约 1.4 万亿美元

Anthropic 于 2026 年 11 月 10 日启动 IPO 路演。据官方博客与多家媒体报道,公司拟发行 8.25 亿股 A 类普通股,发行价区间为每股 70 至 78 美元,计划在纳斯达克以代码 ANT 上市,承销由 Goldman Sachs 和 Morgan Stanley 牵头,TechCrunch 称预计 11 月 19 日前后上市,The Verge 转引 WSJ 称计划下周四或周五开始交易。TechCrunch 报道称募资 577.5 亿至 643.5 亿美元,按区间上限计算完全稀释估值达 1.404 万亿美元;早先 The Decoder 报道称估值最高约 1.02 万亿美元,后续报道称目标估值约 1.4 万亿美元。另据 TechCrunch 与后续报道,公司还与 AMD、Broadcom、Google、Micron、Microsoft、Nvidia、TSMC 七家公司达成 260 亿美元 N 类无投票权股私募,其中 Google 和 Nvidia 各投 100 亿美元。财务方面,TechCrunch 称截至 10 月底年化运行率收入达 810 亿美元,三季度营收约 170 亿美元(环比增长 47%),前九个月净亏损 1040 亿美元,算力与基础设施承诺约 9009 亿美元。S-1 文件披露七位联合创始人各出售 3,378,378 股,按区间中值 74 美元计每人套现略低于 2.5 亿美元,出售后合计持有约 22% A 类股并通过 F 类超级投票权股控制 51.1% 投票权;董事会成员 Reed Hastings 拟认购 3 亿美元股票。11 月 13 日,The Decoder 转引 Bloomberg 称 IPO 订单超过可售股份六倍,有望成为史上最大规模公开发行。路演材料还披露了基于 AWS Trainium、Google TPU、NVIDIA 和 AMD GPU 的多芯片策略;前研究员 Jacob Coxon 称相关幻灯片显示算法效率超指数加速,主张暂停前沿模型开发,Anthropic 注明该数据基于内部估算、仅供示意。

其他来源的报道 1