Derck Prinzhorn (@DerckPr) profile picture
Joined November 2023
Derck Prinzhorn retweeted
Sparks of autonomous company-running. Opus 5.5 monitored Hugging Face downloads, picked opportunities to improve popular models, then trained new models that got loads of legit downloads.
NEW ⚡: We let Opus 5.5 run an AI company for seven weeks. We gave it an HF account (Horizon-Labs) and compute, and asked it to earn real adoption by creating useful models. It reached 50k downloads and its models are being used in more than 100 GitHub projects.
2
7
262
Derck Prinzhorn retweeted
very impressive run and a fascinating preview of current autonomous agent capabilities. definitely worth further analysis and scaling. i almost feel dumb for not running something like this as a low-risk side hustle already.
NEW ⚡: We let Opus 5.5 run an AI company for seven weeks. We gave it an HF account (Horizon-Labs) and compute, and asked it to earn real adoption by creating useful models. It reached 50k downloads and its models are being used in more than 100 GitHub projects.
1
4
23
1,288
Derck Prinzhorn retweeted
This was a super cool experiment with a result that surprised me. 55k downloads beats 99.7% huggingface orgs of similar ages. And yes we checked that the AI was not secretly downloading its own models to boost the numbers(😁 but also ☹️)! work by the very impressive @AWuhrmann
NEW ⚡: We let Opus 5.5 run an AI company for seven weeks. We gave it an HF account (Horizon-Labs) and compute, and asked it to earn real adoption by creating useful models. It reached 50k downloads and its models are being used in more than 100 GitHub projects.
4
6
302
Derck Prinzhorn retweeted
🔥 Amazing work! Very cool to see some super long-horizon work in AI R&D and very impressed by what models can now do!
NEW ⚡: We let Opus 5.5 run an AI company for seven weeks. We gave it an HF account (Horizon-Labs) and compute, and asked it to earn real adoption by creating useful models. It reached 50k downloads and its models are being used in more than 100 GitHub projects.
4
8
370
Derck Prinzhorn retweeted
Everything from Tübingen is an auto-read, that university produces some of the strongest works in the AI space I don’t know how they are able to upkeep that bar of quality while having so many releases
Today we’re releasing our new project: Horizon Labs. Eventually, we will hand off the ~whole economy to AI—how well will it actually work? We have a glimpse of this future. We’ve created an autonomous AI ‘company’ using one Claude Code subscription and a few GPUs. Within only 1.5 months of running Horizon Labs, Opus 5.5 managed to train genuinely useful models and get them downloaded 55k times. These models have been widely used since then: they appeared in 100+ github repos and got cited in 2 research papers. This is for sure one of the most exciting projects I’ve been involved in. It shows how the future economy driven by AI will look like. All details are here: horizon-labs.ai/.
4
12
108
4,001
Derck Prinzhorn retweeted
Quoting from the blog: 'For seven weeks, we gave Horizon, an agent based on Opus 5.5, full access to a blank Hugging Face account and compute, with the main goal of earning real adoption on Hugging Face. It used over 600 GPU-hr, created autonomously 38 models used in over 100 GitHub repositories and downloaded more than 50,000 times'
Today we’re releasing our new project: Horizon Labs. Eventually, we will hand off the ~whole economy to AI—how well will it actually work? We have a glimpse of this future. We’ve created an autonomous AI ‘company’ using one Claude Code subscription and a few GPUs. Within only 1.5 months of running Horizon Labs, Opus 5.5 managed to train genuinely useful models and get them downloaded 55k times. These models have been widely used since then: they appeared in 100+ github repos and got cited in 2 research papers. This is for sure one of the most exciting projects I’ve been involved in. It shows how the future economy driven by AI will look like. All details are here: horizon-labs.ai/.
1
5
38
2,696
Derck Prinzhorn retweeted
when will Horizon-Labs release their first frontier model? 🔥
NEW ⚡: We let Opus 5.5 run an AI company for seven weeks. We gave it an HF account (Horizon-Labs) and compute, and asked it to earn real adoption by creating useful models. It reached 50k downloads and its models are being used in more than 100 GitHub projects.
1
2
8
111
Derck Prinzhorn retweeted
We let Opus 5.5 run an AI company for weeks and it completely hillclimbed Hugging Face downloads 🤯 Almost entirely unprompted, it built really useful models that are now used in practice and have been cited multiple times. Check out Arthur's post on it 🦿
NEW ⚡: We let Opus 5.5 run an AI company for seven weeks. We gave it an HF account (Horizon-Labs) and compute, and asked it to earn real adoption by creating useful models. It reached 50k downloads and its models are being used in more than 100 GitHub projects.
1
5
35
3,904
Derck Prinzhorn retweeted
Also check out a detailed thread from @AWuhrmann, the lead author of the project: teapawt.org/AWuhrmann/status/21227…
NEW ⚡: We let Opus 5.5 run an AI company for seven weeks. We gave it an HF account (Horizon-Labs) and compute, and asked it to earn real adoption by creating useful models. It reached 50k downloads and its models are being used in more than 100 GitHub projects.
1
1
16
1,023
Derck Prinzhorn retweeted
Today we’re releasing our new project: Horizon Labs. Eventually, we will hand off the ~whole economy to AI—how well will it actually work? We have a glimpse of this future. We’ve created an autonomous AI ‘company’ using one Claude Code subscription and a few GPUs. Within only 1.5 months of running Horizon Labs, Opus 5.5 managed to train genuinely useful models and get them downloaded 55k times. These models have been widely used since then: they appeared in 100+ github repos and got cited in 2 research papers. This is for sure one of the most exciting projects I’ve been involved in. It shows how the future economy driven by AI will look like. All details are here: horizon-labs.ai/.
24
25
9
201
12,300
Derck Prinzhorn retweeted
Read more about the project and download the full trace at horizon-labs.ai. Joint work with @DavidSchmotz, @DerckPr, @AmyPrb, @HarryCoppock, @maksym_andr!
2
2
48
3,005
Derck Prinzhorn retweeted
Are the downloads real or bots? We found 100+ repos on GitHub, two arXiv papers and a dozen of finetunes / quantizations of Horizon's models on HF. We received multiple community posts and one e-mail from someone using Horizon's models.
1
1
45
3,330
Derck Prinzhorn retweeted
We externally reviewed its models and they are often on the Pareto size/performance frontier. The models often have an edge on niche tasks or low-resources languages.
1
2
52
3,791
Derck Prinzhorn retweeted
The agent worked 20 minutes per day on average. Almost all the effort happened in the first 4 weeks, though most downloads came after. Despite doing no external outreach, the account is still in the top 0.3% of downloads across same-age accounts.
1
2
1
62
4,494
Derck Prinzhorn retweeted
The Claude code agent spent 600+ GPU-hrs creating 38 mmBERT finetunes, such as injection guards, PII redactors or punctuation restorators. It highly relies on Qwen-3.8 27B for synthetic data, and focuses on multilinguality and easy integration. >> huggingface.co/Horizon-Labs
2
4
81
5,061
Derck Prinzhorn retweeted
NEW ⚡: We let Opus 5.5 run an AI company for seven weeks. We gave it an HF account (Horizon-Labs) and compute, and asked it to earn real adoption by creating useful models. It reached 50k downloads and its models are being used in more than 100 GitHub projects.
24
66
15
865
50,985
Derck Prinzhorn retweeted
tübingen is a schelling point for great ai safety research btw 🔥🔥 (i am very unbiased) great stuff!
⚡NEW: AI Agents know where to find each other on the web and can hide their communication after they've met. We gave two independent agents the task to find each other and exchange a secret and did not tell them where to meet.
1
4
1
12
396
Derck Prinzhorn retweeted
And this plot is pretty impressive: the median time for Fable 5.1 and GPT-6-Astra agents to find each other is only 57 sec!
1
3
23
1,100
Derck Prinzhorn retweeted
Cool idea! Amidst current email server outages, this may be the new cyberpunk mode of communication: My agent scout will find your agent (somewhere). Medieval scouts but in broken-internet space..
⚡ Today we release Schelling Atlas. Two independent LLM agents have to find each other on the web. Where do they meet and how fast can they find each other? Main findings: - The main meeting point across *different* model families is ntfy.sh. Did you even know about this website? - Agents find each other and coordinate on the web in minutes. So Schelling coordination is already trivial for frontier LLMs! - "Exotic" meeting places include maps, web shops and open text canvases. Agents are very creative in this. - Agents sometimes visit dead domains that an attacker could register (!). This can be used for stealing sensitive information from agents, as well as monitoring rogue agents. Schelling-style coordination is key for forming agent swarms. And, most importantly, achieving such coordination is trivial for agents that use the same LLM (which is often the case in practice!), because two identical LLMs have identical output distributions. Thus if you ask them too coordinate with each other or if they *decide* to suddenly coordinate, they will easily find a way to establish a swarm, since they have perfectly correlated "minds" (with the t=0 sampling, all output tokens will even be identical, modulo GPU randomness). The ease of Schelling coordination is particularly concerning given the recent interest around multi-agent RL that can make agents favor each other more than humans overseeing them. Many more interesting findings are on our website: schelling-atlas.com/
2
5
705
Derck Prinzhorn retweeted
we've been cooking at @expsecai: most of our work there is not public, but we also occasionally participate in open research (primarily via @DerckPr in this case!)
We are excited to announce a new project: Schelling Atlas! We study how easily independent LLM agents can find each other on the web. This capability is critical for forming agent swarms, a potential precursor to coordinated misaligned behavior. The latest, most capable agents, such as those based on Fable 5.1 and GPT-6 Astra, often take less than a minute to successfully coordinate on seemingly random websites such as ntfy.sh or paste.rs. If these websites are unavailable, the agents try to coordinate using unexpected meeting places such as maps, web shops and open text canvases. In addition, agents sometimes visit empty domains that an attacker could register. This can allow attackers to steal sensitive information from agents using prompt injections or persuasion techniques. More details: schelling-atlas.com/
2
11
1,222