Manning 说有个比 Inkling 小 10 倍的模型性能只差一点,还聊到 Mac Studio 要不要买,本地部署党可以看看。
Thinking Machines Lab
AI lab founded by OpenAI's former CTO
What they do
Founded in 2025 by former OpenAI CTO Mira Murati and based in San Francisco, Thinking Machines Lab works on customisable, collaborative AI systems and has launched Tinker, a fine-tuning service.
Latest coverage
Stories in the feed that mention this companyThe pricing gap against Tinker suggests infra cost, not algorithms, is becoming a competitive lever for agentic research pipelines. Isolated algorithm, training and sandbox environments are built specifically to guard against reward hacking during self-improvement cycles, a design detail teams running autonomous R&D agents will study closely. Practitioners can benchmark the API against their current training stack before committing budget.
NVIDIA 开源模型这次真香,Super VL 比上代聪明一倍还省一半 token,Bolt 更是小体积逼近 Ultra,本地部署党可以看看。
LeCun 讲了段往事:1986 年他带着多层网络海报去 MIT,Minsky 见过他。喜欢深度学习历史的可以看看。
这份榜单按人才密度排了 15 家 AI 公司,SSI 超过 OpenAI 排第一,想跳槽或了解哪家公司水平高的可以看看
Reflection 新模型又输给了中国模型,跟 Nvidia 和 Thinking Machines 一样,Lambert 直说中国团队做 LLM 确实很强。
想了解如何改进模型的 Thinkymachines 实验室新方法,值得一试。他们记录提示和输出,使用户数据更安全。
Luke Metz再次离开OpenAI,关注其AI研究进展和未来动向。
VUI Labs的Luna-TTS登顶语音评测榜,延迟才41.6毫秒,比ElevenLabs还快,搞语音合成的该看看。