Pinned Tweet
If you have a Macbook Pro M series with at least 64 GB's of RAM you can now run a GPT-4 level LLM locally!
1. Install @ollama
2. Open your terminal and run ollama pull llama3.3
3. Then ollama run llama3.3 "your prompt"
Your own personal AI is here!
Current sub meta tldr:
Anthropic is the best value for dollar
OAI is charging the same while giving you less and their models are not good.
Anthropic Subscriptions Offer 5x+ More Value Than OpenAI
Limit testing every AI subscription plan from Anthropic, OpenAI, Meta, SpaceXAI, MiniMax, Moonshot, Zdotai, Cursor, and Cognition
newsletter.semianalysis.com/…
2
102
opus 5.5 is an amazing model. been mainlining it for work stuff and it breezes through everything i throw at it at any level of the stack.
i think that the regressions we saw between previous models and this one was anthropic trying to make the harness and the model distillation/jailbreak proof and they finally found a way to make it dependable.
3
215
Pretty mediocre announcement list for power users:
1. You are getting fucked if you have pro 20x
2. Your codex threads now will have a “dot” which do exactly the same thing as before but now anthroporphized for normies
3. A new sol model who’s supposedly “better than Astra”
I’m cautiously excited for dev day tomorrow. Their recent releases have been more marketing and hype than actual raw power while also decreasing their plan limits. I hope they release a true opus 5.5/fable challenger and not just a grok bot competitor.
111
Iggy retweeted
Claude can now help you build evaluations and hillclimb on them.
In this article, we share guidance on eval design & skills that Claude Code can use to improve your applications.
claude.dev/blog/automating-e…
I’m cautiously excited for dev day tomorrow. Their recent releases have been more marketing and hype than actual raw power while also decreasing their plan limits. I hope they release a true opus 5.5/fable challenger and not just a grok bot competitor.
Iggy retweeted
GPT-Live-1 is now available in the API.
Bring ChatGPT’s natural back-and-forth to your app, with voice agents that listen while they speak and work with the models and harness you choose.
Buy hard drives, download the best models even if you can’t run them. They will ban everything
This post looks like the start of a VERY sophisticated and well-funded PR operation to get support for Democrats to regulate AI into oblivion. Let me show you how it works:
1.) This guy, with minimal followers and no previous account activity, goes to the Wall Street Journal which publishes an exclusive with quotes from him on his resignation 18 minutes BEFORE this post goes up. Planning was clearly done in advance.
2.) Within hours, it has tens of thousands of reposts and the account has 100k+ followers. The post is punchy, quotable, it almost seems professionally written. The first three accounts to quote tweet it all do so within 15 minutes of the initial posting. Remember, this account had basically zero engagement beforehand, so an organic reach explanation seems unlikely.
According to Grok those accounts are @_NathanCalvin (General Counsel at Encode AI), @peterwildeford (Head of Policy at the AI Policy Network), and @DKokotajlo (Head of the AI Futures Project), all of which are up-and-coming AI-Doomer policy advocacy nonprofits.
The AI Futures Project website says it is funded “primarily” by the Survival and Flourishing Fund, which says on its own website that it has advised Jaan Tallinn, Skype creator and one of the leading investors in Anthropic, to grant over $2.5 million to the AI Futures Project since 2024.
Encode AI says on its website that it is ALSO funded by the Survival and Flourishing Fund, which in turn says that it told Anthropic investor Jaan Tallinn to grant $516,000 to Encode AI in 2025.
And wouldn’t you know it, the Survival and Flourishing Fund ALSO says it told Jaan Tallinn to grant $2 million to the AI Policy Institute, the 501(c)(3) affiliate of the AI Policy Network, as well.
What are the odds that the first three quote tweets of Coxon’s post would all be major AI-restriction policy advocates funded generously by the same donor, who also happens to be one of the leading investors in, and a board member of, Anthropic, the company Coxon was resigning from? And all within 15 minutes of posting (two within ten)?
3.) Jacob Coxon doesn’t have much of a resume, but we do know that, in 2022, he got a $20,159 scholarship for the “long term future scholarship program” from the Good Ventures Foundation, one of the philanthropic vehicles of Dustin Moskovitz, a notorious AI-doomer who has spent tens if not hundreds of millions on policy advocacy to strictly regulate AI, while also being an Anthropic Investor himself.
It also just so happens that the 14th person to quote Coxon’s post was @MaxNadeau_ (27 minutes after posting) who is the program officer for the Technical AI Safety team at Coefficient Giving, another of Moskovitz’s philanthropic spending vehicles. Max is not a frequent poster, his last posts before quoting Coxon were before Labor Day, but he was remarkably quick off the mark for this one.
4.) Basically every major Democrat politician and candidate has suddenly glommed on to this post, and conveniently, as the people cry out foe answers, Bernie Sanders already has a bill written to “ban super intelligence” and regulate AI into oblivion, and will be releasing later this week. The bill, among many other things, will create “a new cabinet-level federal agency to safeguard the public from the dangers of artificial intelligence” that will be “advised by an Artificial Intelligence Advisory Board comprised of experts on artificial intelligence.” Do you think, perhaps, Anthropic and its many investors who fund AI policy advocacy might have interest in getting to place a pet “expert” on the board of an entity that dictates what AI is and isn’t allowed to do? And isn’t it fortuitous that this whistleblower came forward with his oh-so scary stories so close in proximity to the release of the most radical piece of AI legislation ever introduced?
98
Very interesting analysis from @dtzy_88 on the effort levels. I’ve been mainlining light with no issues so far.
Astra 到底开哪个 reasoning effort?
看完 DeepSWE + FrontierCode 1.1,Astra 的 reasoning effort 对位大概是这样:
Astra low/light ≈ GPT-5.6 Sol high
Astra medium ≈ GPT-5.6 Sol max
Astra high ≈ Astra xhigh
也就是说,Astra 基本把上一代 Sol 的 reasoning 档位往前压了两档。
以前 Sol 要开到 max 才能拿到的能力,现在 Astra medium 差不多就能做到。
更有意思的是 high 和 xhigh:
DeepSWE:73% vs 74%
FrontierCode:50.9% vs 50.6%
目前几乎看不到 xhigh 相比 high 有什么稳定优势。
所以如果让我选,Astra high 应该就是目前最舒服的默认档。
至于 max 就很微妙:
DeepSWE:
xhigh $6.52 → max $12.37
成绩反而跌了 74% → 73%
FrontierCode:
xhigh $3.17 → max $4.49
成绩 50.6% → 53.3%
Max 有时候确实能把上限拉高,但有时候花接近 2 倍的钱,成绩反而没涨。
所以 max 更像是“不计成本冲上限”,而不是日常应该常驻的档位。
目前我的开法大概会是:
🪶 轻量任务:low / light
💰 省额度:medium
🍬 Sweet spot:high
🤷 最尴尬:xhigh
🔥 不计成本冲上限:max
简单来说:大多数人直接开 high,可能就是最优解。
你实际用起来怎么样?说说你的看法 👀
1
162
can't believe that in 2020 i still wrote code by hand. spent hours debugging libs i didn't write and monkey patched stuff. read and write code. like a monkey. now i have a monkey of my own. but the monkey has evolved and it doesn't feel like a monkey... most of the time.
89
but if you do this how can i keep rage posting?
We will give one banked reset for every day you don't have access to Astra on your paid ChatGPT plan, starting today. Team is moving mountains to give access as fast as we can.
First one will land in ~ 3 hours. There is still time to create your account if you don't have one.
106
Just got a second claude max sub to work on some hard stuff I was saving up for Astra. Super bummed about this announcement but also according to @every the model is still not as good as fable 5.1 for coding so
¯\_(ツ)_/¯
123
I wonder if Mr Kim Jon Un is piling gpus to run glm 5.3 somewhere in a data center in North Korea to start hacking and stealing at massive levels?
80
Yes, as always vague posting and hype comes before users for OAI
Replying to @synthwavedd
One of the shittiest ways to release a model is to say it’s out, then make users wait a few days to actually get it. What a contradiction.
109
He’s playing 4d chess with branding just to piss on Mr rockets. Now they own the universe codenames while Elon is stuck with “grok”
136
Fable 5.1 limits feel like I’m back on the basic sub while paying for 20x? Can’t wait for Astra
2
124
Daddy chill. I’m barely finishing local benchmarks for glm 5.3 flash on my sparks and now Meta will release more weights? And fable 5.1 is actually good? Astra tomorrow? DADDY CHILLLL
77














