返回
RCreddit.com
21
·10小时前·开发者社区 · RSS

How important is it for Chinese LLMs to reach the Opus 4.8 level?

查看原文
ClaudeQwenNVIDIA模型发布视频生成端侧推理

热度趋势

趋势数据积累中

百分比基于当前可用热度信号,而非评论数或独立用户人数。

推荐理由

Claude 相关模型动态已经出现,适合跟踪能力变化、生态影响和后续可用性。

AI 摘要

根据 Ramp 8 月中旬的数据,Anthropic 最强大的模型 Fable 5 仅占美国公司在其工具上支出的 11%。其余 79% 的支出流向了其他模型,这部分市场价值巨大。随着 Qwen 和 GLM 的新版本发布,中国的 LLM 正接近 Opus 4.8 的水平,并已超越 Sonnet 和其他顶级 LLM。这种快速发展对现有业务构成了真正的威胁,类似于视频游戏行业从全高清向 4K 的转变。当前市场正处于一个狂热阶段,新模型层出不穷,硬件成本不断上升,但从长远来看,AMD 和 Nvidia 等硅芯片供应商有望成为最终的赢家。

In mid-August, Ramp published spending data collected from 70,000 U.S. companies: Fable 5 ,the most powerful and expensive model in Anthropic’s lineup, accounts for just 11% of what those businesses spend on the company’s tools.

The remaining 79% is worth its weight in gold. With the new releases from Qwen and GLM, we are likely close to Opus 4.8, and certainly ahead of Sonnet and the other LLMs shown at the top of the image. The "anti-open-source crusade" therefore comes as no surprise: it is a genuine threat to their business, especially considering the parallel with the video game industry, where the hardware needed to run a game in Full HD became "low-end" within just a few years as 4K took over.

We are in a frenetic phase: new models appearing daily, varying sizes, and—unfortunately—increasingly expensive hardware. But in the long run, I believe the winners will be those selling the silicon for computing (AMD, Nvidia, and soon other competitors) rather than those selling tokens.