Espen retweeted
Over the past week, I’ve been overwhelmed by the love and support of friends, industry colleagues, and people I've never met. Thank you to everyone who has shown kindness to my family and me. It means the world to me.
I came to the United States 53 years ago, and becoming an American citizen is one of the greatest gifts and proudest moments of my life. NVIDIA has been my life’s work. Building it into the company it is today alongside my colleagues has been a privilege and an honor. Nowhere but in America could a company like NVIDIA be possible. I love our country, and I love NVIDIA.
I did not live up to the standards expected of everyone at NVIDIA, and I have made serious mistakes in my personal life. I created problems for the Board that I deeply regret. I own my mistakes completely, and I am committed to cooperating with the Board’s independent investigation.
Since my departure, I’ve seen an outpouring of calls from NVIDIA’s employees, investors, business partners, and customers for me to return as CEO. Many continue to reach out to me directly as the company navigates complex technology, market, competitive, and geopolitical dynamics. NVIDIA is a once-in-a-generation company. If the Board of Directors decides that my return is in the best interest of our shareholders, our employees, the company, and America, I would be honored to answer these calls and return as CEO.
Thank you again to everyone who has shared their good wishes. I am profoundly humbled by your support. To my family: I love you, and I will spend the rest of my life being worthy of yours.
Jensen
Finally Qwen4-Max benchmarks
Qwen4 27B next please
📢 Meet Qwen4-Max, your new AI teammate for work, creativity, and discovery!
Qwen4-Max is built on a new MoE architecture with native vision and a 1M-token context window.
Highlights:
- Outperforms Qwen3.8-Max with up to ~50% fewer tokens, while average cost per task drops over 60% in our internal evaluation.
- Advances real-world productivity through greater proficiency with professional tools, stronger long-horizon execution, and higher-quality deliverables.
- Handles real tasks, including 3D modeling, game development, manufacturing, finance, law, healthcare, and frontier scientific discovery.
- $1.5/M tokens for input, $4.5/M tokens for output, and $0.075/M tokens for implicit caching — all lower than Qwen3.8-Max, with implicit caching down by 70%.
The API is live on QwenCloud. You can also try Qwen4-Max on Qwen Studio, QwenWork and Qoder.
Start building today! 🚀
- Blog: qwen.ai/blog?id=qwen4-max
- API: qwencloud.com/models/qwen4-m…
- Token Plan: qwencloud.com/pricing/token-…
More to come! The open weights of smaller Qwen4 models will be released. Stay tuned. 👀
1
141
Espen retweeted
Replying to @renald_vie
there is no replacing a founder like Jensen.
There never has been, and there can never be, a true succession for a founder like him. Nvidia ends today, and is replaced with a thing cosplaying nvidia. It will be run by custodians trying and failing to preserve the flame he ignited
Espen retweeted
We proudly gave @OpenAI our NVIDIA DGX-1 supercomputer to work on #AI's toughest problems. nvda.ws/2aRtOhE
Couldn't have said it better
For those wondering why I am distraught about Jensen's ouster at Nvidia, let me clear up some details.
First, it's not related to money. I'm doing fine in that department.
Second, it's not related to idol worship, I was actually a critic of many things he did as CEO.
With those out of the way, let me explain. Jensen was a powerful advocate for open models. Or LLMs that you can download and run on your own systems.
He also helped to spearhead Nvidia's Nemotron efforts, which built actual open source models, with full training pipelines, that acted as foundations for future work.
Jensen's advocacy was largely driven out of a combination of passion and necessity in my opinion. The man both believed in open source AI innovation, but also needed it to thrive.
Why did he need it to thrive, you might ask?
Because we live in a world that is increasingly diversifying its AI hardware stack. In the years to come more will come to realize that GPUs are not the end all be all for training or inference. In fact, in some cases it is tantamount to employing a sledgehammer to drive a tack. Excessive and clumsy.
This is why innovations like Google's TPUs gaining PyTorch support is so consequential. As are frontier labs developing their own chips and hardware stacks.
All of that effectively reduces Nvidia's market dominance over time. Jensen understood that. He's a smart man, he could see where this was going.
So he doubled and tripled and even quadrupled down on open weight AI models and open source innovation, seeing it as a way to foam the runway, reducing the hit of the inevitable diversification to Nvidia's margins and revenues.
But he crossed some very powerful forces in the industry along the way, who would much rather reduce the availability of open models, seen as competition, and deliver a much more locked down version of API, where you ultimately pay by the token.
One where self-hosting is restricted, and open models and other innovation is suffocated by regulatory capture.
Jensen was advocating for the Star Trek future. One of abundance, where there was the potential for AI to enhance the lives of so many. Rather than simply a few very big corporate beneficiaries.
Amodei, Altman and their ilk are advocating for the exact opposite. The Star Wars future. One where it is far more likely that the K-shaped socioeconomic distribution of outcomes becomes more stark and dystopian.
After today the Star Wars version of the future became far more likely. This was a huge blow to open models and open source innovation and the reverberations are likely to be felt for years to come.
That is what has me distraught, because as you may know, I am a staunch advocate for a future of abundance, where we can cure diseases and ensure that nobody starves or suffers unnecessarily.
And there you have it. You don't have to agree with my take, as many won't, but you can certainly take the time to understand where I am coming from.
1
206
Espen retweeted
For those wondering why I am distraught about Jensen's ouster at Nvidia, let me clear up some details.
First, it's not related to money. I'm doing fine in that department.
Second, it's not related to idol worship, I was actually a critic of many things he did as CEO.
With those out of the way, let me explain. Jensen was a powerful advocate for open models. Or LLMs that you can download and run on your own systems.
He also helped to spearhead Nvidia's Nemotron efforts, which built actual open source models, with full training pipelines, that acted as foundations for future work.
Jensen's advocacy was largely driven out of a combination of passion and necessity in my opinion. The man both believed in open source AI innovation, but also needed it to thrive.
Why did he need it to thrive, you might ask?
Because we live in a world that is increasingly diversifying its AI hardware stack. In the years to come more will come to realize that GPUs are not the end all be all for training or inference. In fact, in some cases it is tantamount to employing a sledgehammer to drive a tack. Excessive and clumsy.
This is why innovations like Google's TPUs gaining PyTorch support is so consequential. As are frontier labs developing their own chips and hardware stacks.
All of that effectively reduces Nvidia's market dominance over time. Jensen understood that. He's a smart man, he could see where this was going.
So he doubled and tripled and even quadrupled down on open weight AI models and open source innovation, seeing it as a way to foam the runway, reducing the hit of the inevitable diversification to Nvidia's margins and revenues.
But he crossed some very powerful forces in the industry along the way, who would much rather reduce the availability of open models, seen as competition, and deliver a much more locked down version of API, where you ultimately pay by the token.
One where self-hosting is restricted, and open models and other innovation is suffocated by regulatory capture.
Jensen was advocating for the Star Trek future. One of abundance, where there was the potential for AI to enhance the lives of so many. Rather than simply a few very big corporate beneficiaries.
Amodei, Altman and their ilk are advocating for the exact opposite. The Star Wars future. One where it is far more likely that the K-shaped socioeconomic distribution of outcomes becomes more stark and dystopian.
After today the Star Wars version of the future became far more likely. This was a huge blow to open models and open source innovation and the reverberations are likely to be felt for years to come.
That is what has me distraught, because as you may know, I am a staunch advocate for a future of abundance, where we can cure diseases and ensure that nobody starves or suffers unnecessarily.
And there you have it. You don't have to agree with my take, as many won't, but you can certainly take the time to understand where I am coming from.
Espen retweeted
Meet Gemini 4 Hydrogen: built for rapid coding, multimodal synthesis, and everyday workflows.
It finishes tasks 2.5x faster than 3.8 Flash and surpasses Argon on the @ArtificialAnlys Intelligence Index.
Espen retweeted
This whole thing sucks, and I wish none of it had happened. It’s a really sad day!
But I want to make one thing clear: we’ll keep fighting for open-source AI, harder than ever. This is too important to slow down now. Let’s go!
Espen retweeted
For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.
AI will transform every industry, power every company, and be built by every country.
Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.
The world needs both frontier closed models and frontier open models.
images.nvidia.com/pdf/Open-W…
Espen retweeted
NVIDIA announces CEO transition. Read the release: nvidianews.nvidia.com/news/n…
Espen retweeted
In ~8 minutes on a single Colab GPU, I edited Qwen3.5-9B with @GeminiApp Carbon directly in weight space to turn it into a step-by-step math tutor.
Then compiled a RISC-V graphing calculator straight into 11 prepended transformer layers.
Now it's both better at math AND actually shows verified step-by-step work:
• Hard-Math & Graphing: 73.3% -> 96.7% (58/60)
• MATH-500: 56.2% -> 58.2% (15x more verified step-by-step work: 2.8% -> 42.4%)
• GSM8K: 74.3% -> 78.9% (10x verified work: 5.3% -> 53.1%)
• AIME ’25: 16.7% -> 20.0% single-pass
Weights & code: huggingface.co/DJLougen/Qwen…
Espen retweeted
We are looking for a Founding AI Researcher to join our team at @luceboxai
if this is you, reply with your best research work
if you know someone, send them this post
why join:
- probably one of the coolest research jobs right now, making large open models run fast on hardware people can actually own is an amazing mission
- you'll be setting the research agenda with us founders. research is really at our core
- you take open problems from idea to result, your experiments, your call
- post-training and fine-tuning open models end to end on our own hardware
- your research ships, into our open source engine (apache 2.0) and into every lucebox
- if you want to learn where gpu time really goes, you can learn it here, if you already love reading kernel code, you can do it here
- you get one Lucebox maxxed completely for you + you will be working with our A100 on site server.
- backed by Y Combinator, and the founders of Hugging Face, Github, LambdaTest.
what might be dealbreakers to some:
- we work in person in Milan, Italy, so we want you here
- you need to be self-driven, there won’t be a more senior researcher to coach you here
- there is a take-home project
- the italian espresso will ruin all other coffee for you
if you made it this far, reply with your best work, or apply at lucebox.com/careers/founding…. bonus points for submitting a PR to our repo (no slop).
I uploaded my own quant of Ripple 65B last week.
- MXFP8
- 67.8 GB
- Fits single RTX PRO 6000 users
Been working all week on more evals + optimizations. Hopefully done some time during the weekend lol
Also testing it out with the new FLARE inference engine
huggingface.co/EspenJD/Rippl…
3
218
Espen retweeted
ERNIE 5.5 Flash Preview just landed! 🔥
Built on a new architecture designed for efficient inference, the latest open-weight foundation model in the ERNIE family delivers stronger performance in coding, 3D modeling, knowledge work and content creation.
It keeps decoding fast while posting competitive scores on Terminal-Bench, AutomationBench, and other benchmarks.
Here's the rundown:
- 280B total parameters, 14B active per token
- 1M-token context window
- Generation speed above 200 tokens/sec
Model weights will be released by November 25.




























