首页  >>  来自播客: Solving The Money Problem 更新   反馈  

Solving The Money Problem - Grok 4.5 SMOKES Fable 5: Cost & Performance SHOCKER! #shorts

发布时间:   原节目
以下是内容的中文翻译: 视频内容宣布 **xAI 发布了 Grok 4.5**,并强调其相较于 **Fable 5** 等竞争对手具有卓越的性能和显著更低的成本。 以下是所有新闻要点: * **Grok 4.5 发布:**xAI 已在 X 平台正式宣布其 AI 模型 Grok 4.5 的新版本。 * **性能声明:**据称 Grok 4.5 在同类任务上表现优于 **Gemini Ultra** 和 **Claude 3 Opus**。 * **成本优势:**Grok 4.5 每百万 tokens 的定价约为 **1.51 美元**,这与 Fable 5 的每百万 tokens **8.77 美元**形成鲜明对比,且比 Anthropic 的 Opus 每百万 tokens **15 美元**的价格便宜得多。 * **MMLU 分数:**Grok 4.5 在 MMLU(大规模多任务语言理解)基准测试中达到了 **90.2%**,相较于 Grok 1.0 的 86.5% 有了显著提升。 * **GSM8K 分数:**它在 GSM8K(小学数学应用题解决)基准测试中获得了 **90.3%** 的分数。 * **HumanEval 分数:**Grok 4.5 在 HumanEval(代码生成)基准测试中取得了 **88.7%** 的分数。 * **可用性:**Grok 4.5 目前对 **X Premium+ 订阅用户**开放。 * **增强的上下文窗口:**该模型具有 **128,000 tokens** 的扩展上下文窗口,从而支持更长时间的对话和处理更大的文档。 * **多模态能力:**Grok 4.5 引入了**多模态理解**能力,这意味着它可以处理和解释包括**图像、音频和视频**在内的多种形式的数据。 * **具体的多模态用例:**给出的例子包括**描述图像**以及**从截图中生成代码**的能力。 * **整体性能提升:**该模型在**数学、编程和创意写作**方面的能力有所提升。 * **xAI 的愿景:**xAI 旨在让 Grok 成为日常任务的全面“AI 助手”,推动 AI 发展边界,并为**通用人工智能(AGI)**的发展做出贡献。 * **竞争影响:**此次发布被视为加剧了“AI 竞赛”,对 **OpenAI、Google 和 Anthropic** 等主要的 AI 开发者施加了巨大的竞争压力。 * **埃隆·马斯克的角色:**埃隆·马斯克被强调为推动 xAI 创新并挑战现有科技巨头的关键人物。 * **未来影响:**这一发展预示着高级 AI 的可及性可能提高,并可能对 AI 市场造成进一步的颠覆。

The video transcript announces the release of **Grok 4.5 by xAI**, highlighting its superior performance and significantly lower cost compared to competitors like **Fable 5**. Here are all the news items included: * **Grok 4.5 Release:** xAI has officially announced the new version of its AI model, Grok 4.5, on X. * **Performance Claims:** Grok 4.5 is stated to be outperforming **Gemini Ultra** and **Claude 3 Opus** on comparable tasks. * **Cost Advantage:** Grok 4.5 is priced at approximately **$1.51 per 1 million tokens**, a stark contrast to Fable 5's **$8.77 per 1 million tokens**, and significantly cheaper than Anthropic's Opus at **$15 per 1 million tokens**. * **MMLU Score:** Grok 4.5 achieved **90.2% on the MMLU (Massive Multitask Language Understanding) benchmark**, a notable improvement from Grok 1.0's 86.5%. * **GSM8K Score:** It scored **90.3% on GSM8K**, a benchmark for solving grade school math word problems. * **HumanEval Score:** Grok 4.5 achieved **88.7% on HumanEval**, a benchmark for code generation. * **Accessibility:** Grok 4.5 is currently available to **X Premium+ subscribers**. * **Enhanced Context Window:** The model features an expanded context window of **128,000 tokens**, allowing for much longer conversations and processing of larger documents. * **Multi-modal Capabilities:** Grok 4.5 introduces **multi-modal understanding**, meaning it can process and interpret various forms of data including **images, audio, and video**. * **Specific Multi-modal Use Cases:** Examples given include the ability to **describe an image** and **generate code from a screenshot**. * **Improved General Performance:** The model is noted for improved capabilities in **math, coding, and creative writing**. * **xAI's Vision:** xAI aims for Grok to serve as a comprehensive "AI assistant" for daily tasks, pushing AI boundaries, and contributing to the development of **Artificial General Intelligence (AGI)**. * **Competitive Impact:** The release is seen as intensifying the "AI race," putting significant competitive pressure on major AI developers like **OpenAI, Google, and Anthropic**. * **Elon Musk's Role:** Elon Musk is highlighted as a key figure driving xAI's innovation and challenging established tech giants. * **Future Implications:** The development suggests a potential for wider accessibility to advanced AI and further disruption in the AI market.