此刻温度最高

48 小时内每个独立来源只算一次，热度每 24 小时减半。排在前面的，是真正有很多人在说的 10 件事。

2026·10·14

NO.01↓ 16%

Google 在全球发布 Gemini 4 Argon，扩大访问范围并报告其前 72 小时处理了 25 万亿个 token

Google launched Gemini 4 Argon globally through paid developer services and subscriptions, expanded access through Antigravity and third-party apps, and processed about 25 trillion tokens in its first 72 hours, according to Google’s Logan Kilpatrick. Google offers Argon through the paid Gemini API and Gemini platform, and to Pro and Ultra subscribers in the Gemini app and Search. Antigravity launched with Ultra access and is adding paid Pro subscribers through its desktop ap…

24 小时热度峰值91 · 14 小时前

24 小时前现在

最新进展In just the first 72 hours since Gemini 4 Argon launched, we processed ~25 trillion tokens, the demand for intelligence is incredible and still under appreciated!

Simon Willison、宝玉 · X 等 9 个来源·11位参与者

热度指数

NO.02↓ 16%

Meta 发布 Muse Wonder 模型，基准测试成绩提升但 Muse Code 使用量统计存在错误

Meta released Muse Wonder through the Meta Model API and Muse Code and Muse Wonder Full for Muse Code subscribers, while Alexandr Wang said the team was addressing a bug in how both models counted against subscribers’ usage limits. Wang promised resets once the bug was fixed; his subsequent update said issues remained. TechCrunch reports that Wonder’s API pricing is $4.00 per million input tokens and $20.00 per million output tokens. Artificial Analysis reports that Wonder a…

9个来源 · 10位参与者

NO.03发酵中↑ 51%

Anthropic 测试 AI 在安卓设备上的复制能力，并警告未来可能出现物联网蠕虫

Anthropic’s Frontier Red Team used a remotely running GLM-5.2 to deploy Llama3.1-8B across simulated Android devices, but the experiments did not demonstrate a fully autonomous AI worm. In an isolated network of eight GCP nodes running Android 15 emulators, GLM-5.2 reached all eight nodes in four of five trials and seven in the remaining trial. The setup provided unauthenticated Android Debug Bridge access and omitted actual hacking, limiting what it establishes about spreadi…

4个来源 · 4位参与者

继续看No.04–10

参与者 · 24 小时走势 · 热度指数

04
涉嫌录制会议室内容的 OpenAI 前工程师被下令羁押候审发酵中
A US court ordered former OpenAI engineer Abdulrahman Alfozan to remain detained pending trial on a charge of using or attempting to use a device to intercept oral communications, after OpenAI found a recording device in an executive conference room and fired him. Magistrate Judge Alex G. Tse ordered his continued detention after prosecutors cited alleged Saudi connections and government-funded travel as evidence of flight risk. Prosecutors acknowledged that Alfozan may have …
3个来源
21↑ 50%
05
Hugging Face 推广推测解码直播，vLLM 鼓励观众收看新
Hugging Face shared a livestream about speeding up local large language models with speculative decoding, and vLLM encouraged its audience to watch. Hugging Face also posted an X broadcast titled “Local AI 202: Get Massive Speed-ups with Speculative Decoding.” Hugging Face described the session as explaining how to achieve multiplicative speedups for local LLMs. vLLM called it a practical look at the technique and said making speculative decoding easier to use benefits the i…
2个来源
19新上榜
06
Anthropic 随代码变更激增，将其单体仓库从 Git 迁移到 Scion
Anthropic built Scion, its own version control system, and fully migrated its monorepo from Git in five months to address infrastructure bottlenecks as daily merged code changes rose from around 2,500 in May to more than 30,000 in September. Anthropic said seven people built the system, with its repository approaching 100 GB of main-branch history and CI performing roughly 25 million checkouts a day. Scion retrieves files on demand, moves operations to the server, and uses l…
3个来源
18↓ 16%
07
Cursor 加入 X 订阅，Grok 用量共享，第三方模型单独计费
Cursor announced that it has joined The X Subscription, bringing it into a bundle with Grok and X Premium while allowing existing Cursor subscribers to keep their current plans or switch. Cursor said paid plans share one usage pool across Cursor, Grok Build, and other Grok products and provide twice the coding usage on Grok models as the equivalent Cursor plan. On the new plans, third-party models in Cursor move to API pricing. Baoyu explained that these plans include no thi…
3个来源
11↓ 16%
08
Meta推出“你的算法”，让用户自定义Facebook动态和Reels推荐
Meta launched Your Algorithm on Facebook, giving users a way to describe in natural language what they want to see more or less of in Feed and Reels. The feature, which Meta first introduced on Instagram, lets Facebook users add and manage topics they want recommended. An updated three-dot menu provides access to Your Algorithm and the “Interested” and “Not interested” options on posts. Users may receive follow-up questions to clarify their preferences. Meta says natural lan…
2个来源
11↓ 16%
09
Thinking Machines 修复 Muse 配额问题，并宣布重置 Muse Code
Thinking Machines fixed the issue causing unusually fast quota usage for Muse Wonder and Muse Wonder Full, Alexandr Wang said, announcing that users would now see much lower usage against their limits and that Muse Code had been reset as promised after the fix. Wang later said the team had been fixing issues with the model rollout over time and thanked users for giving it another try. The specific quota fix applies to Muse Wonder and Muse Wonder Full; the reset announcement …
2个来源
10↓ 16%
10
Character.AI申请破产保护，迪士尼出价2000万美元收购其资产并向员工提供职位
Character.AI has filed for Chapter 11 bankruptcy and agreed to sell some assets and a non-exclusive perpetual technology license to Disney for $20 million, with the sale still open to higher bids. Disney also agreed to offer jobs to at least 60 employees, a majority of Character.AI’s workforce. A legal filing cited by Bloomberg estimated their annual compensation at $60 million and said the vast majority of those offered jobs had accepted. The bankruptcy followed lawsuits an…
2个来源
10↓ 16%

热度是怎么算的？了解榜单

热度来自参与同一事件的独立账号与机构，重复采集只算一次，并按 24 小时半衰期衰减。它衡量讨论活跃程度，不是报道质量评分。

榜单统计过去 48 小时。趋势只比较持续覆盖的同一组信源；它反映我们的监测范围，不代表全网人数。缺少可比历史时，不展示趋势线。

信源名单只展示可公开阅读的报道来源；讨论参与者还包括只计入热度的账号与机构。同一机构的多个渠道可能合并计数，因此参与者不一定多于信源数。点击事件可查看各方报道与观点。

爆: 讨论快速增加
新: 首报 6 小时内
发酵中: 讨论仍在增加

此刻温度最高

继续看No.04–10

涉嫌录制会议室内容的 OpenAI 前工程师被下令羁押候审发酵中

Hugging Face 推广推测解码直播，vLLM 鼓励观众收看新