We are also announcing 100% cashback but also on Higgsfield API through Comfy Router Get your API key from Higgsfield, put into Comfy Router, enjoy the cashback! VCs are funding free model access. You should take advantage of it :) And congrats to Higgsfield on reaching a $1B run rate limited to Higgsfield models, unused cashback expires on September 30.

伙伴们在尝试让 AI 改网页的时候，有没有在这一步卡住： “把按钮挪到标题下面，去把这块删掉，还要旁边那张卡突出一点。” 看着页面觉得很清楚，但是写成文字，却得去解释到底是去搞哪个按钮、哪张卡。能不能直接在截图上圈出来，让 AI 照着改？ 我用 @AntLingAGI Ling-3.0-flash-VL 做了个小测试。准备了一张活动报名页，同时把源码和标注截图一起交给它： 红圈里的横幅删掉，按钮移到蓝箭头指向的位置，黄框里的票种设为唯一推荐项。 同一个任务测了三次，三次首版都把按钮放错了。 接着，我把前两次实际渲染出来的页面截图回传给它，让它继续修改。两次都在一轮反馈后通过了全部 17 项验收。第三次停在首版，保留为失败；这组最终完成了 2/3。 这里模型负责看图、修改代码，本地工具负责运行页面、截图和验收。 如果你也在用 AI 改落地页、活动页，这次测试有个值得试的方法： 先在截图上标明要改哪里；首版不对，就把实际页面再截给它看，指出没完成的地方。修改前也写清哪些内容要保留，避免挪好了按钮，却改掉价格或弄坏交互。 这也是我为什么设了 17 项检查：除了位置和推荐卡，还要核对原有文字，点开、关闭报名弹窗，展开 FAQ，再看手机端有没有遮挡、溢出和控制台报错。页面看着改好了，还得点一遍才知道能不能用。 不过，换个目标，它仍然会出错。我把黄框改到“观察席”，它却继续推荐“共创席”。这个变体只测了一次，失败了。 所以，在这组样本里，我看到了模型收到页面截图后修正代码的效果；但换一个位置、换一张卡，仍然需要重新检查。还不能把整页修改放心交给它后就不管了。 57 秒视频里放了标注输入、反馈后的变化、成功和失败结果，标注为已生成结果回放。 你可以照这个过程试自己的页面，也能看到哪些地方需要留给人验收。 - 蚂蚁数科国际站 ：computrix.ai/sg/models/model… - 蚂蚁数科 maas 国内站：maas.antdigital.com/models/m… - Ling studio：chat.ant-ling.com/chat 这几个都可以，目前可用的体验地址

71.6% of small business owners have filmed a video for their business and never posted it. 53.9% said it didn't look professional enough. 48.6% hated how they looked. We asked 1,000+ owners what happens when an AI avatar shows up for them. Full report: heygen.com/the-state-of-ai-a…

We’ve just released the preview version of our latest paper: Infinite-Parameter LLMs — Generating and Adapting Weights from Live Data. In it, we demonstrate that large language models can learn up to 1,000x faster through Bayesian Self-learning transformers (BAST) than SOTA training algorithms engineered by human researchers. This breakthrough marks a critical shift in AI research, from cost-intensive, static-weight AI to energy-efficient, dynamic-weight AI. For a decade, building more advanced models has been achieved by training bigger ones on more data. This approach is now hitting a ceiling with the supply of pretraining text projected to run out in the next two to five years. Meanwhile, the fast adoption of AI agents is producing an unprecedented amount of continuously-growing data that models can learn from. None of it is captured, locked in individual sessions and lost as soon as the agent completes the task. Bringing self-learning AI agents to users captures the value of that data. We validated that BAST LLMs overcome the fundamental limitations of memory-based learning, such as Retrieval-Augmented Generation (RAG), in both cost and performance, across long-context conversations. Without underlying architectural innovation like BAST, RAG, larger context windows, and agent scaffolding suffer a drop in performance and escalating input token cost due to the static weight of LLMs. This research underscores a fundamental premise of our work at Boltzbit to achieve General Learning Intelligence and democratise model ownership. Scaling up static-weight model size or layering on more workarounds cannot address the spiraling cost of training AI systems. The viable path is a new AI architecture that adapts its own weights. Preview version of the paper: arxiv.org/pdf/2609.18842