AI - 分类 - Simi Studio /categories/ai/ AI - 分类 - Simi Studio Hugo -- gohugo.iozh-CNsimi@simi.studio (Simi) simi@simi.studio (Simi)Wed, 15 Apr 2026 10:00:00 +0800 Claude Mythos：Anthropic 发布了一个模型，然后把它藏起来 /posts/claude-mythos-preview/ Wed, 15 Apr 2026 10:00:00 +0800 simi@simi.studio (Simi) /posts/claude-mythos-preview/ Anthropic 发布 Claude Mythos。

然后把它藏起来了。

不是没发布。是发布了，但普通人用不了。

]]> GPT-6 预训练完成：18 个月、10 万张 H100、200 万 token 上下文 /posts/gpt-6-pretraining-complete/ Tue, 14 Apr 2026 17:00:00 +0800 simi@simi.studio (Simi) /posts/gpt-6-pretraining-complete/ AI Agent 失控：从模型崇拜到环境工程 /posts/harness-engineering/ Tue, 14 Apr 2026 10:00:00 +0800 simi@simi.studio (Simi) /posts/harness-engineering/ AI Agent 安全红线：我的生产环境检查清单 /posts/practical-ai-security/ Fri, 03 Apr 2026 10:30:00 +0800 simi@simi.studio (Simi) /posts/practical-ai-security/ Qwen3.5-Omni：阿里在 215 项音视频任务上超越了 Gemini-3.1 Pro /posts/qwen-3-5-omni-multimodal/ Mon, 30 Mar 2026 10:00:00 +0800 simi@simi.studio (Simi) /posts/qwen-3-5-omni-multimodal/ Claude Code 最佳实践 /posts/claude-code-best-practices/ Sun, 29 Mar 2026 10:00:00 +0800 simi@simi.studio (Simi) /posts/claude-code-best-practices/ Edge AI 部署：让 LLM 跑在你的设备上 /posts/edge-ai-deployment/ Thu, 26 Mar 2026 14:00:00 +0800 simi@simi.studio (Simi) /posts/edge-ai-deployment/ 垂直模型崛起：代码模型、语音模型、图像模型的专业分工 /posts/specialized-ai-models/ Sat, 21 Mar 2026 10:00:00 +0800 simi@simi.studio (Simi) /posts/specialized-ai-models/ 多模态 LLM 成熟了吗：2026 年初的实测 /posts/multimodal-llm-evolution/ Tue, 10 Mar 2026 14:30:00 +0800 simi@simi.studio (Simi) /posts/multimodal-llm-evolution/ GPT-5.4 的百万 token 上下文：终于不再截断 /posts/gpt-5-4-million-token-context/ Thu, 05 Mar 2026 10:00:00 +0800 simi@simi.studio (Simi) /posts/gpt-5-4-million-token-context/