Eira @eiralune26 - Twitter Profile

震惊，一个 9B 的国产开源小模型，把 Opus 4.5 按在地上摩擦了 Coinbase 拿阿里的 Qwen3.5-9B，用自家数据做强化学习微调，专门抓买币欺诈在他们自己的 16,140 笔交易考卷上，四项检测指标全赢 Opus 4.5，F1 高了 9.6 个百分点线上中位延迟 0.683 秒，Opus 4.5 要 1.515 秒训练只用一张显卡，花了不到 100 美元更离谱的是，Coinbase 自己测出来，新出的 Opus 5、Sonnet 5、GPT-5.6 抓欺诈的召回和 F1 都不如各自的上一代连 CEO 本人都出来说，以后公司里会有越来越多专科小模型在窄活上干翻全能大模型考卷是 Coinbase 自己出的，外人没法复核，但不到 100 美元换一个专科选手，这买卖太划算了🤯 https://t.co/qKlrvG6tpd

Brian Armstrong

@brian_armstrong

2 days ago

We were able to fine tune an open weight model (Qwen3.5-9B) on proprietary internal data at @coinbase, and it out performed frontier models, at much lower cost and speed, for a specialized task (fraud prevention in this case). I think we'll see more of this pattern inside companies. The frontier models are amazing generalists. But increasingly you'll see specialist models out perform on more narrow tasks, either working on teams, or by themselves. We're making it easy to fine tune models internally for our employees. We even have a skill for it in our internal harness that makes it easy to spin these up and start training them: "/fine-tune-and-serve-llm I want to fine-tune Qwen for payments risk review. help me get started" I imagine we'll have some backend services that could be replaced entirely by this new architecture.

brian_armstrong's tweet photo. We were able to fine tune an open weight model (Qwen3.5-9B) on proprietary internal data at @coinbase, and it out performed frontier models, at much lower cost and speed, for a specialized task (fraud prevention in this case).

I think we'll see more of this pattern inside companies. The frontier models are amazing generalists. But increasingly you'll see specialist models out perform on more narrow tasks, either working on teams, or by themselves.

We're making it easy to fine tune models internally for our employees. We even have a skill for it in our internal harness that makes it easy to spin these up and start training them:

"/fine-tune-and-serve-llm I want to fine-tune Qwen for payments risk review. help me get started"

I imagine we'll have some backend services that could be replaced entirely by this new architecture.

597

9K

776

6K

1M

0

28

Eira

@EiraLune26

about 18 hours ago

上周二 OpenAI 一口气往 GitHub 倒了 722 篇 AI 写的数学论文，第二天就撤了 3 篇😅 撤的理由特别数学：一篇里有个符号错了，另外两篇是搭在它上面的，一起塌而且 722 是篇数不是题数。一个还没发布的内部模型被丢了大约 4000 道题，只挑出够分量的写成论文数学与 AI 咨询小组 AGMAI 9 月底的建议里说得很直白：AI 现在能写出连提问的人都看不懂、核不了、也担不了责的证明我的担心是，机器出论文的速度，人可能已经核不��来了。我猜接下来数学圈最抢手的是审稿人

0

19