Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)

Just completed GPU kernel bench, interesting reveals
- DeepSeek v4.1 flash > glm-5.3 flash
- DeepSeek v4.1 flash > haiku 5.5
- hy4 preview == gemini 3.8 flash
- Qwen 27B FP8 > Q4 Strata Next version @LottoLabs

🚨重磅爆料｜国内AI赛道又迎来关键节点！月之暗面的Kimi K3.1已经敲定10月中上旬上线。重点打磨长程Agent能力。

月之暗面两套模型同步推进，不再死磕加大参数量。他们用电脑操作记录+真实任务数据做训练，专门解决长流程任务容易断档的痛点。

团队还在打磨5-6T的超大模型，但最大难题不是算法，高端训练算力供给跟不上。

现在采用大小模型搭配的玩法：旗舰模型冲硬核能力，轻量K2.8负责编程、日常任务，一边跑一边收集真实场景数据。国产芯片跑推理没问题，但是做大模型预训练，短期很难替代英伟达。

Deepseek v4.1-flash is insane.

Going back to other models feels way slower and discouraging.

v4.1-flash is below GLM-5.3-flash for quality, but i'm still sticking with v4.1-flash more because it can build and iterate way faster lol.

We gave Fable 5 and GPT-5.6 Sol 3000 GPU-hours and tasked them with developing a novel post-training technique that would improve on a standard pre-existing baseline (GRPO).

There is only one princeling in the politburo – Xi Jinping himself.

There are no ladies… sometimes, you jess have to accept when somfin is based as fuq.

I remember how circa 2016, russia was that funny Cyka Blyat meme country. You saw a lot of jokes and memes about shirtless putin, hardbass, vodka, Cheeki Breeki, and foul-mouthed CS:GO gamers. russia blew a lot of cultural capital with the full-scale invasion of Ukraine.

这位朋友，让我们面对这个问题：
巴西的生育系数是1.55，而2.1以上，才可以实现人口更替；
LGBTQ人士中，这个平均系数则为0.1
这意味着什么？
你们冲在了巴西抵抗的最前沿，但20年，30年后，你们老去之后呢？
不再有人，就什么也没有，彻底失败了。
这就是为什么，中国禁止在公共层面大力宣传这个。

必须先有人。
如果停止产生后代，自己消灭自己，那无论你们抗议的怎么英勇，事实上是做了你们反对的人最想做却做不到的事情——解决反对的人。
并且，这件事情是你们主动帮对手做的。
当然，我们也面临巨大的生育危机问题，我们也做错了事，不过这是另一个故事了，我们正在努力振作，解决这个问题。
我以上说的这些，或许有些冒犯，但很抱歉，事情就是这样。

Business Insider is reporting that Google is internally testing a new Gemini 4 checkpoint named Carbon that matches Opus 5.5 in coding.

Inference scaling works, but every agent eventually faces the same problem: its context window fills up.

In this essay, we argue the only scalable fix leads straight to swarms.

> none have displayed any hint of instrumental convergence (generalized power-seeking and self-preservation)

This is also clearly overstated! Agents will absolutely figure out how to get money, broader internet access, etc. instrumentally while working on fairly minor tasks.

From Egan Jones. Since 2004, the Chinese economy has outgrown the US by basically any metric you can think of. But American financial assets have clobbered Chinese ones however you measure it.

Scoop: Top executives at Anthropic, OpenAI and other AI companies are privately gaming out scenarios for a public and political revolt after a catastrophic AI event…They assume one in ‘27…
axios.com/2026/10/09/ai-…

'I had assumed that this problem was analytically intractable. After working independently for 100 minutes with Max thinking, GPT-6 Astra produced the initial derivation, yielding a self-consistent problem whose solution matched simulations of large networks.

I found this derivation intriguing, albeit complicated and somewhat mysterious (see Al Methodology in the paper). After investigating the structure of the proposed solution more deeply, together with Al models, I arrived at a physically transparent derivation.'

'Many people in the theoretical sciences are anxious about Al. However, in my field at least, we are nowhere close to fully understanding the high-dimensional computations unfolding in our brains. With Al, we could reach a level of understanding that so far we have only gestured toward. So, I am excited.'

Semianalysis said 92% API margin, and about -360% in subscription when maxing out the whole month on opus 5.5

Axios is reporting that the top labs are wargaming scenarios for the aftermath of a catastrophic event in the next 6 - 12 months, that some executives now see as being inevitable.

A note from our research leaders:

Last week we parted ways with Jasmine, Mikita, and Tomek after a thorough investigation found they violated clear policies on handling sensitive information. Our internal investigation uncovered a significant breach of trust beyond what’s outlined in the letter they published and we stand by the decision to not continue their employment. We generally keep individual employment matters private and don't believe a back and forth would be productive or lead to a resolution, but we want to address the points they raised in their letter directly.

- We want to be very clear that these decisions were not about raising safety concerns or speaking out. Safety and research debates happen every day at OpenAI, often spirited and highly critical. We actively encourage these discussions and consider them essential to making the right decisions. We cannot do the work in front of us without a high degree of trust. We will continue to be extremely forgiving of our team making good-faith mistakes. We have not and do not terminate any of our employees for raising concerns.

- We are actively finalizing contracts with third-party safety assessors and will announce details in the coming weeks. People across the company have been working really hard on getting these partnerships up and running. We are committed to embedding external assessors and continue to make close collaboration with independent safety organizations a core part of our safety work. Many of our researchers already work with 3p safety organizations productively.

- We agree with the letter that preserving the monitorability of frontier models requires an industry-wide commitment, including from OpenAI. Monitorability has long been a core piece of our research program, and something we continue to invest significant resources in (see our publications on Monitoring Monitorability and the subsequent open sourcing of monitorability evals, our system card for GPT-6 Astra, Jakub’s blog and post on X, and the numerous blog posts on our Alignment blog on the topic).

We are deeply sad about this outcome. We appreciated Jasmine, Mikita, and Tomek’s contributions to AI safety at OpenAI and their willingness to speak up and challenge ideas. We championed their voices, supported their work, and placed enormous trust in them. These decisions were not about them raising safety concerns. We have always encouraged that and always will.

I realize @AcerFur is being ironic, but the popularity of the idea that AI "merely explores the convex hull" mystifies me. Why wouldn't it be more like the subgroup generated by existing mathematical knowledge? (That's also a metaphor -- I'm not saying there's a group structure.)

as these potential hacks are well discussed in their tech report, it is a bit hard to believe at first that they left the answer recoverable in 2/3 of the coding tasks.

but after going back through their tech report, there might really be a few things that can be improved in their RL environments.

The first is how they removed the reference fix. Their tech report says each task repo is set back to right before the fix, and the later commits are removed. However, according to the audit, only the branches got deleted. And in Git a branch is just a name, so the commit with the full fix was still sitting in the repo's storage. The MiMo team's own check only looks at the names, so it came back clean when the repo was not. Deleting the leftover commits and having the check look at the storage itself would close this.

The second is the "below 2%" hack rate from their tech report. That is the share of training runs where a grader caught the model copying an answer. Runs that copied without getting caught just counted as passes. So a clearer explanation of how the grader audits each run would help explain the gap between the 2% hack rate and the 2/3 leak rate.

The third is the cleanup list. Files touched by the fix had a later "last modified" time than the rest of the repo, and the audit shows the model using that to figure out where to edit. Timestamps are not on their cleanup list, so adding them would help.

claude pro订阅的opus5.5换算使用单价和deepseek v4.1 flash其实是一样的，更何况你还要搭配claude code使用，这进一步提高了claude的token效率使得deepseek相比之下看起来性价比更低了。
我是这样想的