埃隆·马斯克的xAI在Grok-2中展示了增强的功能和图像生成能力

币界网Publicado em 2024-08-14Última atualização em 2024-08-14

币界网报道：

埃隆·马斯克的人工智能初创公司xAI推出了其聊天机器人Grok-2的最新版本。它宣布，新的人工智能助手的测试模式将面向付费X用户（以前是推特）。

xAI于8月13日向公众发布了Grok-2。它宣传其最新的人工智能模型是4月份发布的Grok 1.5版本的改进版本。新模型有两个大型语言模型（LLM）版本：Grok-2和功能较弱的Grok-2 mini。

xAI称其Grok-2为最重要的聊天机器人模型

最新版本是xAI聊天机器人自2023年11月发布Grok-1以来的第三次迭代。该公司将此次发布描述为对其前身的重大改进。

独立评论对Grok-2、ChatGPT-4o和Gemini 1.5 pro的比较结果喜忧参半。该比较由Chatbot Arena完成，这是一个评估大型模型系统组织（LMSYS Org）创建的LLM的平台

Grok 2对竞争对手的排名（来源：xAI）

根据LMSYS组织的说法，Grok-2在某些方面与GPT-4o相匹配，在编码、数学和硬提示方面表现出色。然而，它在所有人工智能功能中排名第三，低于GPT-4o和Gemini 1.5。

沃顿商学院教授Ethan Mollick也承认Grok-2的能力，但他指出，没有任何人工智能模型能超越OpenAI的GPT-4。他说：

“现在有五种GPT-4级型号：GPT-4o、Claude 3.5、Gemini 1.5、Llama 3.1和现在的Grok 2。所有实验室都表示，还有继续大幅改进的空间，但我们还没有看到任何模型真正超越GPT-4……”

新的聊天机器人目前仅供X Premium和Premium+用户使用。然而，这家初创公司计划在8月底之前向开发人员提供其应用程序编程接口（API）。

Grok用户现在可以使用它来生成图像

Grok-2版本的一个显著改进是生成图像的能力。与Midjouney和DALL-E等其他AI图像生成器相比，它对生成的图像的限制更少。用户已经用它来创建政治主题的图像。

生成图像的能力有助于Grok-2与其他聊天机器人进行比较，因为几乎所有顶级模型，如OpenAI的ChatGPT、Google Gemini和Meta的Llama，都可以创建图像。马斯克甚至转发了Grok根据提示生成乔治·华盛顿肖像的截图。

Leituras Relacionadas

Not Just DeepSeek, Big Tech Companies Want to 'Abandon' NVIDIA

The article discusses how major tech companies are attempting to reduce their reliance on Nvidia, despite its dominant position in the AI chip market, where it enjoys a 75.2% GAAP gross margin. Companies like DeepSeek are adapting their models to run on domestic alternatives such as Huawei’s Ascend chips, while in the U.S., Google and Meta are developing their own AI chips (TPU and MTIA series) to complement external partnerships. Nvidia’s CEO, Jensen Huang, acknowledges that Moore’s Law is fading and that export restrictions may slow China’s AI development but could ultimately spur a self-sufficient ecosystem. Notably, Chinese firms are leading in open-source models, which could eventually challenge Nvidia’s monopoly. OpenAI is actively diversifying away from Nvidia, signing a $20 billion deal with Cerebras, a startup using wafer-scale chips to reduce latency and cost. Cerebras, founded by Andrew Feldman, aims to challenge Nvidia with its unique architecture but faces financial and competitive risks, including heavy dependence on OpenAI and geopolitical tensions. While competition is intensifying with players like AMD and Groq (which partnered with Nvidia), the overall demand for compute continues to grow. The market is shifting toward a diversified supplier model, though Nvidia remains a formidable force.

marsbitHá 20m

Not Just DeepSeek, Big Tech Companies Want to 'Abandon' NVIDIA

marsbitHá 20m

Ozak AI’s Forecast Range Expands to $8–$11 as AI Adoption Accelerates Across DeFi, Data, and Automation Markets

Ozak AI ($OZ) is emerging as a leading AI-backed crypto investment, positioned at the intersection of AI, DePIN (Decentralized Physical Infrastructure Network), and tokenized economic growth. With AI adoption accelerating in DeFi, data analytics, and automation, the project's forecasted value has expanded to a range of $8–$11. Currently in advanced presale stages, Ozak AI has raised over $6 million, with tokens priced at $0.014, indicating strong pre-listing commitment. Its DePIN architecture supports scalable, decentralized AI processing, distinguishing it from other AI tokens. The $OZ token offers utility through staking, voting, and network expansion, driving demand. The $8–$11 valuation is based on successful presale activity, exchange listings, and growing adoption in decentralized markets, positioning Ozak AI for significant potential growth.

TheNewsCryptoHá 31m

Ozak AI’s Forecast Range Expands to $8–$11 as AI Adoption Accelerates Across DeFi, Data, and Automation Markets

TheNewsCryptoHá 31m

Illustrating the Capital Market After DeepSeek V4's Launch: Zhipu and MiniMax Plunge, NVIDIA Panics

DeepSeek V4, a 1T parameter MoE model with a 285B Flash version, has been fully open-sourced under Apache 2.0, triggering significant reactions across capital markets. Chinese AI chipmakers like Cambricon and Hygon saw major stock gains, with Cambricon rising 60% monthly. In contrast, Hong Kong-listed AI firms Zhipu and MiniMax dropped over 7%, facing heavy short-selling. NVIDIA’s shares dipped, with analysts noting a "decoupling" of Chinese and North American AI inference demand. The launch intensified competition in the AI model space, following 11 major releases in 30 days, including GPT-5.5 and Llama 4. Unlike others, V4’s permissive licensing and full open-source release challenged closed-source models on performance, cost, and accessibility. Critically, V4 announced Day-0 support for domestic chips like Huawei’s Ascend 950PR and Cambricon’s Siyuan 590, offering better cost-performance than NVIDIA counterparts. This shift reduces reliance on CUDA, aligning with NVIDIA CEO’s earlier concerns about Chinese AI chips threatening its dominance. The move signals a tangible step in China’s AI supply chain independence, redirecting compute demand to local manufacturers like Hua Hong Semiconductor.

marsbitHá 1h

Illustrating the Capital Market After DeepSeek V4's Launch: Zhipu and MiniMax Plunge, NVIDIA Panics

marsbitHá 1h

Crypto Coalition Urges Senate To Fast-Track CLARITY Act As US Leadership Faces Critical Moment

A coalition of over 120 crypto industry organizations, including the Crypto Council for Innovation and the Blockchain Association, is urging the U.S. Senate Banking Committee to fast-track the CLARITY Act, a comprehensive market structure bill for digital assets. They argue that timely legislation is critical to ensure consumer protection, clarify regulatory roles, and maintain U.S. leadership in financial innovation, warning that delay could cede advantages to other jurisdictions. The push comes amid reports of a potential delay until mid-May due to banking sector opposition, particularly concerning restrictions on stablecoin yields. Industry leaders emphasize that May is a crucial window for action before political attention shifts to election campaigns.

bitcoinistHá 1h

Crypto Coalition Urges Senate To Fast-Track CLARITY Act As US Leadership Faces Critical Moment

bitcoinistHá 1h

Day 6 of the rsETH Incident: DeFi United Secures Approximately $100 Million in Intentional Commitments, but a $50 Million Gap Remains

On April 18, Kelp DAO’s rsETH LayerZero bridge was exploited, resulting in the unauthorized minting of 116.5k rsETH (approx. $292M). The attacker borrowed around $190M on Aave V3. The Arbitrum Security Council froze 30,766 ETH linked to the incident. DeFi United, a cross-protocol rescue initiative led by Awe, was formed to cover a total shortfall of 112.2k rsETH ($258M). As of April 24, several protocols have pledged around $100M in support, though most commitments are still under DAO voting or discussion. Key pledges include: - Golem: 1,000 ETH ($2.3M) - Aave founder Stani Kulechov: 5,000 ETH ($11.5M) - EtherFi: up to 5,000 ETH ($11.5M) - Lido: up to 2,500 stETH ($5.75M), contingent on full coverage - Mantle: proposed a $69M loan to Aave DAO under specific terms The remaining shortfall is estimated at $50M. Aave’s treasury and safety module (~$236M combined) can cover the worst-case bad debt scenario ($230M). Three potential loss distribution paths were outlined by DefiLlama’s 0xngmi: 1. Uniform 18.5% haircut for all rsETH holders: Aave bad debt ~$216M 2. Only protect Mainnet, abandon L2: bad debt up to $341M 3. Repay only pre-attack holders: technically difficult, ~$91M net loss KelpDAO has not yet announced a specific plan. The success of DeFi United depends heavily on KelpDAO’s final decision on loss allocation.

marsbitHá 1h

Day 6 of the rsETH Incident: DeFi United Secures Approximately $100 Million in Intentional Commitments, but a $50 Million Gap Remains

marsbitHá 1h

Trading

Spot
Futuros
活动图片

Categorias popularesright-arrow

Etiquetas Popularesright-arrow