-
1
Opus 5.5发布:降价+安全升级
Anthropic称Opus 5.5是其测试过最强模型,同步降价,并针对网络安全风险新增防护机制
Anthropic releases Opus 5.5 with lower prices and Fable-level performance ↗4.6 -
2
OpenAI发布GPT-6双模型Sol/Luna
OpenAI同时推出两款新模型,主打更低成本与更少错误,覆盖日常办公不同性价比场景
OpenAI launches GPT-6 Sol and Luna, boasting lower cost and fewer mistakes ↗4.6 -
3
Snorkel AI估值翻三倍至35亿美元
这家数据标注公司完成3.5亿美元E轮融资,估值三倍增至35亿美元,受AI训练数据需求拉动
Snorkel AI triples valuation to $3.5B as demand for AI training data booms ↗3.4 -
4
微软联合业界铲除AI诈骗平台
AI客服诈骗平台EvilTokens曾攻陷1.2万个微软账号,微软周二宣布联合行业将其打掉
Microsoft disrupts AI-assisted platform that compromised 12,000 ↗3.4 -
5
Rabbit转型,推云端Agent系统OS3
Rabbit推出跨平台云端AI agent「OS3」,不再需要R1硬件即可使用,摆脱专属设备依赖
Rabbit's new AI agent doesn't need an R1 to run ↗3.0 -
6
骁龙8至尊版发布,本地跑300亿参数模型
高通发布骁龙8至尊Gen 6及Extreme版芯片,顶配机型可在本地运行300亿参数MoE模型
Qualcomm launches two new smartphone chips with emphasis on AI ↗3.0 -
7
数学争议后,OpenAI请外部专家把关
此前数学基准测试成绩引发信誉危机,OpenAI周一宣布成立独立小组咨询顶尖数学家
OpenAI wants to consult elite mathematicians about how to not fumble again ↗3.0 -
8
Meta承认Muse抄袭OpenClaw设计
Meta承认其AI助手Muse并非从零打造,深受OpenClaw启发,连工作区文件命名都相似
Meta admits Muse's likeness to OpenClaw isn't a coincidence ↗3.0 -
9
2.5
-
10
a16z办AI学院,携手谷歌/Meta等
风投a16z推出面向年轻人的创业培训「学院」,首批10家合作方含Palantir、谷歌与Meta
Andreessen Horowitz is launching an 'academy' with no homework and partnerships with Palantir, Google, and Meta ↗2.4 -
11
英国AISI联手HF推基准可复现
英国AI安全研究院(AISI)与EvalEval合作,推动AI基准测试结果标准化、可复现
How UK AISI and EvalEval Are Making Benchmark Results Reproducible ↗2.3 -
12
AstroForge用AI自主指挥探测器
AstroForge下一代探测器Autonomy-1将由小型Transformer模型自主掌控飞行任务
AstroForge is putting AI in command of its next spacecraft ↗2.1