https://guijiagi.com/tags/%E5%A4%9A%E6%A8%A1%E6%80%81/Recent content in 多模态 on 硅基 AGI · 智能体学习与测评Hugozh-cn本站内容采用 CC BY-NC-SA 4.0 国际许可协议授权Thu, 16 Jul 2026 11:09:00 +0800https://guijiagi.com/posts/b1-4a19c4f9/Thu, 16 Jul 2026 11:09:00 +0800https://guijiagi.com/posts/b1-4a19c4f9/系统讲解多模态大模型的技术演进，从CLIP双塔到GPT-4V统一架构再到原生多模态训练https://guijiagi.com/posts/b2-19902ccc/Thu, 16 Jul 2026 10:07:00 +0800https://guijiagi.com/posts/b2-19902ccc/梳理多模态大模型从早期CLIP双塔架构到GPT-4o原生多模态的技术演进路径，分析架构设计选择与工程权衡https://guijiagi.com/posts/article-95/Mon, 13 Jul 2026 08:40:00 +0800https://guijiagi.com/posts/article-95/梳理AI从纯文本到图文音视频多模态感知的演进历程，解析关键技术突破与未来方向https://guijiagi.com/posts/article-09/Sun, 12 Jul 2026 18:20:00 +0800https://guijiagi.com/posts/article-09/探讨多模态AI Agent的架构设计原则，从感知融合到跨模态推理的完整技术栈https://guijiagi.com/posts/b2-ad8b68bd/Sun, 12 Jul 2026 18:20:00 +0800https://guijiagi.com/posts/b2-ad8b68bd/探讨多模态AI Agent的架构设计原则，从感知融合到跨模态推理的完整技术栈https://guijiagi.com/posts/seedance-25-video-era/Tue, 07 Jul 2026 13:30:00 +0800https://guijiagi.com/posts/seedance-25-video-era/字节豆包视频生成模型Seedance 2.5支持30秒单段视频直出，50个多模态素材联合输入，第三方测评超越GPT-5.6https://guijiagi.com/posts/multimodal-model-2026-guide/Thu, 02 Jul 2026 10:58:00 +0800https://guijiagi.com/posts/multimodal-model-2026-guide/2026年多模态大模型全面选型指南，覆盖图文、视频、音频融合理解https://guijiagi.com/posts/vision-model-2026-selection/Thu, 02 Jul 2026 10:54:00 +0800https://guijiagi.com/posts/vision-model-2026-selection/2026年视觉语言模型全面选型指南，覆盖OCR、图表理解、视频分析等场景https://guijiagi.com/posts/openai-realtime-api-voice/Thu, 02 Jul 2026 10:19:00 +0800https://guijiagi.com/posts/openai-realtime-api-voice/OpenAI发布实时语音API，延迟低至200ms，语音交互体验大幅提升https://guijiagi.com/posts/baidu-ernie-5-release/Thu, 02 Jul 2026 10:14:00 +0800https://guijiagi.com/posts/baidu-ernie-5-release/百度文心一言5.0发布，多模态推理能力实现重大突破https://guijiagi.com/posts/google-gemini-3-ultra-review/Thu, 02 Jul 2026 10:03:00 +0800https://guijiagi.com/posts/google-gemini-3-ultra-review/Google Gemini 3 Ultra全面评测：原生多模态是否真正领先https://guijiagi.com/posts/agent-ux-design-2026/Tue, 30 Jun 2026 17:00:00 +0800https://guijiagi.com/posts/agent-ux-design-2026/2026年AI Agent UX设计指南：对话设计、反馈机制、错误处理、多模态交互、信任建立等核心设计原则https://guijiagi.com/posts/multimodal-vision-benchmark-2026/Tue, 30 Jun 2026 09:10:00 +0800https://guijiagi.com/posts/multimodal-vision-benchmark-2026/2026年多模态大模型深度评测：GPT-5 Vision、Claude 5 Vision、Gemini 3 Pro、Qwen-VL Max在视觉理解、OCR、视频分析等维度的全面对比https://guijiagi.com/posts/ai-video-generation-2026-landscape/Sun, 28 Jun 2026 12:00:00 +0800https://guijiagi.com/posts/ai-video-generation-2026-landscape/2026年AI视频生成领域四大主流模型的全景对比，从画质、时长、一致性到商业化的深度分析https://guijiagi.com/posts/multimodal-agent-practice/Sun, 28 Jun 2026 12:00:00 +0800https://guijiagi.com/posts/multimodal-agent-practice/手把手教你构建多模态Agent，实现图像理解、语音交互、视频分析等能力，包含完整代码示例和架构设计https://guijiagi.com/posts/multimodal-fusion-architectures/Sun, 28 Jun 2026 11:13:00 +0800https://guijiagi.com/posts/multimodal-fusion-architectures/深入对比多模态大模型的三种融合架构：Early Fusion、Late Fusion、Cross-Attention的原理与2026年实践https://guijiagi.com/posts/hermes-agent-architecture/Sun, 28 Jun 2026 11:00:00 +0800https://guijiagi.com/posts/hermes-agent-architecture/深度解析 Hermes Agent 爱马仕智能体的技术架构设计、核心组件、推理引擎与应用实践https://guijiagi.com/posts/2026-mid-year-ai-industry-report-five-key-trends/Sun, 28 Jun 2026 10:00:00 +0800https://guijiagi.com/posts/2026-mid-year-ai-industry-report-five-key-trends/2026 年上半年 AI 行业全景扫描：从 Agent 商业化、多模态融合、推理成本下降、AI 科学发现到监管落地，深度解析五大核心趋势https://guijiagi.com/posts/multimodal-rag-image-text-hybrid-retrieval/Sun, 28 Jun 2026 10:00:00 +0800https://guijiagi.com/posts/multimodal-rag-image-text-hybrid-retrieval/深入多模态 RAG 的工程实现，涵盖图文统一编码、混合检索策略、跨模态重排序等核心技术https://guijiagi.com/posts/multimodal-vision-models-2026-comprehensive-evaluation/Sun, 28 Jun 2026 10:00:00 +0800https://guijiagi.com/posts/multimodal-vision-models-2026-comprehensive-evaluation/2026年主流多模态大模型视觉理解能力全面横评，覆盖图像理解、视频理解、OCR与空间推理https://guijiagi.com/posts/multimodal-agent-architecture/Sat, 27 Jun 2026 15:00:00 +0800https://guijiagi.com/posts/multimodal-agent-architecture/多模态Agent架构设计https://guijiagi.com/posts/multimodal-rag-practice/Sat, 27 Jun 2026 15:00:00 +0800https://guijiagi.com/posts/multimodal-rag-practice/多模态RAG架构实践https://guijiagi.com/posts/multimodal-eval-methodology/Sat, 27 Jun 2026 15:00:00 +0800https://guijiagi.com/posts/multimodal-eval-methodology/多模态模型评测方法论https://guijiagi.com/posts/multimodal-model-selection-guide/Sat, 27 Jun 2026 15:00:00 +0800https://guijiagi.com/posts/multimodal-model-selection-guide/多模态模型选型指南https://guijiagi.com/posts/multimodal-fusion-architecture-2026/Sat, 27 Jun 2026 15:00:00 +0800https://guijiagi.com/posts/multimodal-fusion-architecture-2026/深度解析多模态融合架构原理，从早期融合到晚期融合再到混合架构的技术演进https://guijiagi.com/posts/multimodal-fusion-architecture/Sat, 27 Jun 2026 15:00:00 +0800https://guijiagi.com/posts/multimodal-fusion-architecture/多模态融合架构原理深度解析https://guijiagi.com/posts/gpt-55-%E6%B7%B1%E5%BA%A6%E8%AF%84%E6%B5%8B-%E5%A4%9A%E6%A8%A1%E6%80%81%E6%8E%A8%E7%90%86%E7%9A%84%E6%96%B0%E6%A0%87%E6%9D%86/Sat, 27 Jun 2026 00:00:00 +0800https://guijiagi.com/posts/gpt-55-%E6%B7%B1%E5%BA%A6%E8%AF%84%E6%B5%8B-%E5%A4%9A%E6%A8%A1%E6%80%81%E6%8E%A8%E7%90%86%E7%9A%84%E6%96%B0%E6%A0%87%E6%9D%86/全面评测 GPT-5.5 的推理能力、多模态表现与 Agent 构建能力，对比 GPT-5 提升幅度，分析其在实际应用中的表现https://guijiagi.com/posts/multimodal-agent-design/Fri, 26 Jun 2026 14:00:00 +0800https://guijiagi.com/posts/multimodal-agent-design/探讨多模态智能体的架构设计，涵盖视觉语言模型、语音交互、视频理解与跨模态推理，附完整系统架构与代码实现。https://guijiagi.com/posts/multimodal-model-selection/Thu, 25 Jun 2026 12:00:00 +0800https://guijiagi.com/posts/multimodal-model-selection/全面对比 2026 年主流多模态大模型：GPT-5o、Gemini 2.0 Pro、LLaVA-1.6、Qwen-VL2、InternVL3 在图像、视频、文档理解上的表现与选型建议。https://guijiagi.com/posts/multimodal-eval-method/Thu, 25 Jun 2026 12:00:00 +0800https://guijiagi.com/posts/multimodal-eval-method/系统介绍多模态大模型评估方法，涵盖视觉理解、跨模态推理、图文一致性、OCR、视频理解等核心维度的评估框架与实现。https://guijiagi.com/posts/gemini-3-5-flash-review/Thu, 25 Jun 2026 10:00:00 +0800https://guijiagi.com/posts/gemini-3-5-flash-review/Google 发布 Gemini 3.5 Flash，首 token 延迟 280ms，原生多模态生成，深度集成搜索，Agent 场景全面覆盖https://guijiagi.com/posts/multimodal-eval-2026/Thu, 25 Jun 2026 10:00:00 +0800https://guijiagi.com/posts/multimodal-eval-2026/2026 年多模态模型评估全景：视觉、音频、视频理解与跨模态推理https://guijiagi.com/posts/vision-language-model-guide/Thu, 25 Jun 2026 10:00:00 +0800https://guijiagi.com/posts/vision-language-model-guide/主流视觉语言模型深度对比，涵盖架构原理、OCR/图表/视频理解能力与成本延迟分析https://guijiagi.com/posts/gemini-40-%E9%A2%84%E5%91%8A-%E8%B0%B7%E6%AD%8C%E7%9A%84%E5%85%A8%E6%A8%A1%E6%80%81%E9%87%8E%E5%BF%83/Thu, 25 Jun 2026 00:00:00 +0800https://guijiagi.com/posts/gemini-40-%E9%A2%84%E5%91%8A-%E8%B0%B7%E6%AD%8C%E7%9A%84%E5%85%A8%E6%A8%A1%E6%80%81%E9%87%8E%E5%BF%83/Google DeepMind 预告 Gemini 4.0，支持文本/图像/视频/音频/代码原生多模态，分析其技术路线与对行业的影响https://guijiagi.com/posts/multimodal-eval-guide/Wed, 24 Jun 2026 16:00:00 +0800https://guijiagi.com/posts/multimodal-eval-guide/从感知到推理到生成，系统性评估视觉语言模型（VLM）的完整框架https://guijiagi.com/posts/vision-model-selection/Wed, 24 Jun 2026 15:00:00 +0800https://guijiagi.com/posts/vision-model-selection/全面解析 VLM 架构演进、主流模型对比、OCR 与视频理解能力、部署成本与选型决策。https://guijiagi.com/posts/embodied-ai-2026/Tue, 23 Jun 2026 15:30:00 +0800https://guijiagi.com/posts/embodied-ai-2026/2026年具身智能领域的技术突破、代表产品和未来展望