Feed

only in SF

america and china are adversaries

america has china beat on war, and china has america beat on labor.

they’re each just trying to close the gap.

china is making weapons because they already have workers.

america is making workers because we already have weapons.

---

why are you as a man riding an escalator when there’s perfectly good stairs next to it…have you no shame? no pride?

“oooh machine please carry me i am so weak and fragile im scared of a little pe=mgh i cant ascend these small obstacles”

you are little boy. a frail little boy!

RT HDP
build the mana fortress .. as ordained within the river of time

---

RT bebeal
hackerman 9000 reporting in

Just got access to a new model and honestly, I can't believe it's performance. I trained it myself and it's kind of dog shit I really thought this experiment would work

RT Nathan Lambert
Over the weekend I had Codex parse 500K arXiv AI/ML papers since ChatGPT to understand which open models are used for research.

In 2024, ~30% of papers mentioned an American open model and only 10% a Chinese model.

Today, ~40% of papers mention a Chinese (open) LLM, and only 25-30% an American one. Chinese models are the default for research. Chinese mentions are still growing while American open models are stagnating.

When looking at this data it's important to remember that papers substantially lag model releases, as research takes a long time. Qwen's steady growth is reflective of this, but so is Llama's lasting power.

Some more observations:

1. Qwen has been steadily growing, and today 1/3 of papers which mention any LLM mention qwen. OpenAI's closed models are the highest overall, at ~37%.

2. Llama peaked around April of 2025 at 30% of papers which mention any LLM (including ChatGPT etc). Llama 4 was released at about the same time, and Llama has been declining since.

3. Gemini and Claude are less common than the leading open models, mentioned in 10-15% of papers puts them behind all of Qwen, Llama, and DeepSeek. Open models should be and are the foundations of open research.

The % of papers mentioning any LLM have been steadily climbing since 2023.
| Year | January | April | July | October |
| 2023 | 10.43% | 15.39% | 18.69% | 32.18% |
| 2024 | 29.70% | 33.93% | 35.70% | 44.25% |
| 2025 | 39.23% | 45.28% | 44.94% | 53.52% |
| 2026 | 55.49% | 57.26% | 53.14% | TBD

Now over 50% of AI papers, from 10% in 2023.

Other notes:
- Gemma and Mistral hover around 5-10%.
- Our beloved fully-open Olmo models have been ~1% since the first release in Jan. 2024.
- DeepSeek has a clear jump after R1 in Jan. 2025
- Data derived from the most popular ML arXiv categories: cs. AI, cs. CL, cs. CV, cs. LG, stat. ML

Just like our downloads and derivative model data, this is updated daily on the Interconnects Open Model Dashboard.

I HAVE HAD TOO MUCH CAFFEINE AAAAAAAAAAAAAAAAAAAAAAAAAAAA

install linux now

I had deepseek download love island episodes (legally of course haha) and then it put all of the episodes as shortcuts on the steam machine's desktop

In my wife's words "I love this thing" (the steam machine)

I gave k-stack an email and then gave my dad the email so now he can send work to it with zipped files (mostly processing latex for the courses he teaches) and it has its own little sandbox and happily does the work for him

RT daniel debow
think less about trump

think more about selling, building, competing & growing our companies, products and services.

I don’t care about having a ton of money.

But sometimes i dream about owning a big mahogany desk with a vintage globe and an emeralite bankers lamp. Victorian oak walls covered in lacquer book shelves, black on the outside and ruby on the inside. A leather recliner and a shisha on the side table. A smoky haze filling my lungs, and then the room, and then dissolving into brass ceiling lights. A fish tank in the far corner, not because I like fish but because they force me to enter my cove everyday, no matter how strenuous life gets.

And while i’m down there i might as well leaf through a few pages of Shel Silverstein or Ray Bradbury. Maybe I’ll linger for a while, and remind myself that I used to dream of this.

I don’t care about having a ton of money, but if i did, that’s what i would spend it on.

my curiosity exceeds your will to live

RT NEOMECHANICA

need to start an open source telepathy benchmark

are you ready?

RT hayden
how it feels to open freecad for the first time

Airpods are probably the greatest product of all time

It’s just flawless execution across the entire stack

- Digital signal processing- their active noise cancellation is flawless, live translation shows promise.

- Mechanical design: universal fit, even without soft tips. extremely miniaturized, low weight, and mass manufacturable.

- Marketing: people hated this concept when it came out and now everyone has them.

- UX: very satisfying opening and closing haptics, great charging experience, seamless bluetooth connect/disconnect. on-stem controls.

and somehow they keep improving with every generation?!

“Being robust to a single failure is obviously important, but equally as important is observability of those failures.”

read this. then read it again. then read it a third time.

getting the “D” score wrong in your FMEA’s RPN is one of the most common mistakes engineers make.

---

An alcoholic can still curse the bottle.

You can call me weak but you can’t call me wrong.

---

RT Michael Morelli
My dog ChooChoo before BPC-157 vs. after BPC-157.

---

if a drone this loud and obnoxious comes anywhere near my house i will shoot it down idgaf

very anti-human design. you get your package faster but at the cost of the serenity of your neighbors.

---