有人又捕捉到 OpenAI 传闻中的 GPT-Aeon 痕迹了，想追 OpenAI 新模型动态的可以先蹲一下。
Andrew Curran 在 X 上发布了一条关于 GPT-Aeon 的目击消息，并附上链接。GPT-Aeon 此前尚未被 OpenAI 官方确认，属于传闻中的新模型名称。这条推文本身仅包含一个链接，未给出基准分数或发布时间等细节。
GPT-Aeon sighting. https://t.co/kD9iUgn4wM
有人从 Codex 统计和云端目录里挖出 OpenAI 和 Anthropic 各一个未发布模型，喜欢追模型动态的可以先看看线索。
The teaser suggests Cohere is shifting toward open weights, a notable move for a company long focused on enterprise closed deployments. Developers and enterprises evaluating open-source alternatives can check the blog link for specifications as they are published. Watch for how these models compare with Llama and Qwen on enterprise RAG workloads.
The practical lesson is that compounding many small, accuracy-checked optimizations can deliver a 10x throughput gain without new hardware, and the details are documented in a public blog post. Inference engineers and teams serving document pipelines can study the writeup to cut their own GPU costs. Since some improvements will land in open-source vLLM, the gains will eventually benefit anyone serving models, not just LlamaParse users.
DeepSeek 团队搞了个 Super-agent 原型，能自己盯着聊天群和 GitHub 接活干活，不用人派任务了。
Mistral 又上新了，Mistral Large 4 开放权重 1.05T，上下文窗口拉到 1M，已接入 Open Math OS 供获批团队直接用。
