Horizon created 38 models
For seven weeks, the Horizon agent based on Opus 5.5 autonomously created 38 models used in over 100 GitHub repositories and downloaded more than 50,000 times. It processed approximately 1.9 billion tokens using 600 GPU-hours of compute.
- Field
- applying language models
- What they did
- The AI agent Horizon used 600 GPU hours to autonomously create 38 models that were downloaded over 50,000 times and used in 100 GitHub repositories.
- Why it matters
- This allows for testing and deploying models in real-world projects without human intervention, saving time and resources.
#horizon#opus 5.5#autonomous agent#hugging face#llm



