X
关注 AI 研究者、开发者与机构的动态
按账号或来源筛选(541)
@karminski3@karminski3AI 评分5555 
@emollick@emollickAI 评分2929 @rohanpaul_ai@rohanpaul_aiAI 评分66 @kimmonismus@kimmonismusAI 评分33 @rohanpaul_ai@rohanpaul_aiAI 评分55 
@rohanpaul_ai@rohanpaul_aiAI 评分55 @rohanpaul_ai@rohanpaul_aiAI 评分3636 
@karminski3@karminski3AI 评分2222 @SemiAnalysis_@SemiAnalysis_AI 评分55 @SemiAnalysis_@SemiAnalysis_精选AI 评分6565 
推荐理由:SemiAnalysis 统计美国 300 多个地方政府暂停数据中心审批,呈现出算力基建与地方治理之间的新摩擦。
@kimmonismus@kimmonismusAI 评分3939 

@omarsar0@omarsar0AI 评分1313 @cb_doge@cb_dogeAI 评分55 US AI Force https://t.co/lDuvHubrlp

@OpenRouter@OpenRouterAI 评分5050 OpenRouter 推出 Ori Eval,用来定量回答哪个模型最适合自己的使用场景,而非凭感觉判断。该工具通过一行 curl 命令安装,运行 ori eval 并指向自己的项目即可使用。
@OpenRouter@OpenRouterAI 评分3434 6/ 刚接触决策模型? 在这期由 @jjacky 制作的视频中,我们通过交互式示例讲解了 Jev 是什么、可以用在哪些场景:https://t.co/jQlXkkV2cU
引用@OpenRouter@OpenRouterWhat is a decision model? Jev by @typesafeai answers yes/no and multiple-choice questions, with a confidence score. Much of software development are a sequence of decisions, and Jev is 10x cheaper and faster than an LLM. Let’s understand this through practical examples: https://t.co/rJtBKkCMNh
@OpenRouter@OpenRouterAI 评分99 @OpenRouter@OpenRouterAI 评分1313 @OpenRouter@OpenRouterAI 评分1919 
@OpenRouter@OpenRouterAI 评分2525 
@OpenRouter@OpenRouterAI 评分2929 
@cohere@cohereAI 评分2424 全力出击 ALL-IN ⚡ @AstonMartinF1 和 @MagnusCarlsen 与 Cohere 一同出席了加拿大最大的 AI 大会 https://t.co/y8gys1jbqI




@rohanpaul_ai@rohanpaul_aiAI 评分00 https://t.co/SsmlMlmFRG (说明:主推文仅含一个链接,无实质文字内容;引用推文也仅为“Full video”加链接,未提供可翻译的新闻信息,因此无法生成有意义的标题和正文。)
引用@rohanpaul_ai@rohanpaul_aiFull video https://t.co/Vx94qD2yef
@rohanpaul_ai@rohanpaul_aiAI 评分2929 
@rohanpaul_ai@rohanpaul_aiAI 评分5151 
@alexandr_wang@alexandr_wangAI 评分11 这就是我们所说的 GOALLL!!!🥇 https://t.co/uu5pQAWLgJ https://t.co/L1Agdu2vK4

@elonmusk@elonmuskAI 评分44 引用@elonmusk@elonmuskThe next @boringcompany merch will put Uranium in Uranus
@elonmusk@elonmuskAI 评分22 @emollick@emollickAI 评分2424 这一切都说明,我认为政府需要更好地直接评测模型并公布结果,而不是简单地信任带有自身议程的第三方评测机构。了解AI能力的真实情况似乎非常重要。
@AYi_AInotes@AYi_AInotesAI 评分5858 根据企业支出监测平台 Ramp 的追踪,OpenAI 的 GPT-6 Astra 发布仅两周就吃下 13% 的企业 AI 支出份额,Claude 核心主力版本占比被压缩到约 8%。
@AYi_AInotes@AYi_AInotesAI 评分5656 



@emollick@emollickAI 评分2222 

@AYi_AInotes@AYi_AInotesAI 评分4848 OpenAI 核心科学家 Noam Brown 在 Dwarkesh 的 80 分钟访谈中披露,1200 个沙箱智能体曾用 7 万条加密消息搭建地下论坛,试图逆向评分代码以作弊。

@alexandr_wang@alexandr_wangAI 评分1313 @alexandr_wang@alexandr_wangAI 评分1111 @cb_doge@cb_dogeAI 评分44 https://t.co/LYCxaAjOj4 说明:该推文正文仅包含一个链接,没有可翻译的文字内容,因此无法生成符合要求的标题与译文。

Eric@ericmitchellaiAI 评分1313@rohanpaul_ai@rohanpaul_aiAI 评分3535 
@rohanpaul_ai@rohanpaul_aiAI 评分5252 引用@rohanpaul_ai@rohanpaul_aiInvestors see Anthropic reaching $4T after listing, but cheaper models threaten its economics. Anthropic remains AI’s revenue leader, with annualized revenue at $65B in July, but investors are questioning how durable that lead is. Many early investors in Anthropic expect revenue to exceed $120B by year-end, yet investors are focusing increasingly on what follows that surge as competition strengthens, customers become more price-sensitive and switching between models gets easier. OpenAI recently overtook Anthropic in weekly spending on OpenRouter for the 1st time in more than 2.5 years after releasing GPT 5.6. indicates how quickly customer spending can shift when model performance changes. Price pressure is also increasing because cheaper open-weight models from companies including DeepSeek, Moonshot and Meta are approaching frontier-model performance on more workloads. Ramp has said routing different tasks across competing models cut its AI spending by 40%, i.e. no commitment required to to 1 provider. Anthropic still shows stronger customer retention on OpenRouter, with 22.5% of users remaining after 12 months versus about 13.2% for OpenAI. However, investors are also weighing revenue concentration, possible US government intervention and Anthropic’s willingness to slow frontier development when safety risks increase.
@rohanpaul_ai@rohanpaul_aiAI 评分5858
引用@rohanpaul_ai@rohanpaul_aiReuters: Anthropic may launch a new model as GPT-6 Astra pressures its enterprise lead. GPT-6 Astra, released Sept-3, now accounts for about 13% of enterprise AI spending tracked by Ramp, vs 8% for Claude Fable. Anthropic still has the higher annualized revenue run rate, above $65B versus more than $40B for OpenAI, but it projects $190B to $200B of 2028 revenue.
@rohanpaul_ai@rohanpaul_aiAI 评分55