Joined May 2009
- Tweets 3,175
- Following 494
- Followers 12,461
- Likes 2,485
123 Photos and videos
Pinned Tweet
4 Dec 2025
Excited to start as a Founding Data Scientist at @prior_labs working together to revolutionize tabular data science.
We are going to push the boundaries of Tabular Foundation Models with TabPFN and change how people approach tabular data. Let's go!
13
6
136
16,881
May 15
Opus is so much better than Codex on autoresearch, not even a competition at this point
May 14
Automating AI research is the next major step in AI
We let Claude Code (Opus 4.7) and Codex (GPT 5.5) run autonomously on the nanoGPT speedrun optimizer track using our idle compute. ~10k runs, ~14k H200 hours
Opus now holds the record at 2930 steps vs the 2990 human baseline
1
28
4,968
May 13
This is a big nerf to programmatic use with subscription. 200$ per month is basically one 5 hour session
May 13
Monthly credit amounts vary by plan:
Pro: $20
Max 5x: $100
Max 20x: $200
Team Standard: $20/seat
Team Premium: $100/seat
Enterprise: Varies by seat type
After you claim the credit, it resets with each billing cycle. Credits do not rollover.
4
2,767
May 12
A new leap forward for tabular foundation models: TabPFN-3 is now live scaling accuracy, context size, inference performance and capabilities across the board. No tuning required, just load the model, pass your data and get state of the art predictions.
2
3
57
4,452
May 7
I wish Claude Code had a certain amount of jokers each week that allow to reset the five hour usage window. That would spread out the usage also a bit because everyone runs into limits at different times usually. @bcherny @amorriscode
5
761
Philipp Singer retweeted
May 4
Huge news: @prior_labs has signed a definitive agreement to be acquired by @SAP.
€1B invested over four years to build a globally-leading frontier AI lab for structured data — in Europe, in the open.
Independent entity. Same team, same mission, same open models. A massive boost to what we can do. The mission just got accelerated.
Founders’ statement: priorlabs.ai/blog-posts/prio…
(Deal subject to regulatory approval; terms not disclosed.)
36
32
510
51,733
Apr 7
It is funny that codex cli is basically unsupported for Ubuntu 24 and noone fixes it for weeks now.
1
6
1,667
Apr 1
Kind of funny that April 1st in 2026 feels like any other day, you don't know what is real and fake anyways most of the time.
1
6
761
Mar 26
We just released TabPFN 2.6 @prior_labs being the new frontier on TabArena. It improves across tasks and is an early look of where we are heading next. You can find the weights and more details on HF huggingface.co/Prior-Labs/ta…
1
17
56
4,495
Mar 24
Timeline is so bad again, it was a bit better the last few weeks but this week Im getting so much trash and gossip again.
1
561
Mar 24
Possible future, but why via GUI and not everything programmatically?
Mar 23
The future where I never have to open up my laptop to get work done is becoming real very fast
1
2
1
1,008
Mar 19
This is such a big pain point where Im surprised they haven't tackled this yet. We desperately need some form of account-level configuration for skills, mcps, settings that sync across devices and between desktop and CC @bcherny
Mar 19
anthropic should add a simple feature to sync skills between claude chat, claude cowork and claude code and between teams
i see how much people are struggling with this
1
5
1,377
Mar 17
People love to reset sessions, me included, but sometimes I end up in these god sessions when using coding agents. The context gets longer and longer, it keeps compacting, but the quality just is always there and I never want this session to go away.
10
38
3,064
Mar 15
I am somewhat surprised that Latex Beamer slides do not play a bigger role in the current agentic knowledge work stack and that we are still building pptx files.
2
9
1,451
Mar 13
Okay using transformers.js with whisper tiny locally on my Pixel 10 feels pretty nice. I wonder why not more apps add support for local speech to text directly on device.
4
944
Mar 12
Tinkering with building my own agent and development orchestration setup with Proxmox. Customizing this will be more and more valuable over time I believe.
12
1,412
Mar 11
So many things currently remind me about the good old gaming days where you never wanted to log out of WoW in case the login servers go down.
We're experiencing issues with Claude Code and Claude.ai where some users may not be able to log in & others may experience slower than usual performance.
We're on it and working hard to bring service back to normal, thanks for bearing with us.
5
1,382
Mar 11
To be fair I asked Codex yesterday to review and judge a Claude based repo and it told me "this is well designed with high taste".
Asked Claude 4.6 Opus to review a big refactor from Codex (GPT 5.4 xhigh) - "No issues found, ready to commit".
Maybe because I didn't tell Claude the work was done by Codex 🤪
2
1,355
Mar 10
Every couple of weeks I try Codex again because of my bubble here, and it keeps disappointing me. I have to be doing something wrong. But it keeps going off rails and just does things buggy or things I did not ask about. Maybe the app is much better than the cli.
7
16
3,520
Mar 6
I still don't understand how to use Codex apart from implementing clear features purely focusing on coding. It works for 10 minutes and then it gives me the shortest bullet points list of findings back and I have to prompt it three more times to get more details out of it.
3
7
3,540