AutoScientist
#8 in Wetenschop- & Bio-KIadaption · siet 2026-05-13 · 12× · tolest 30. Juni 2026
24
Momentum
AutoScientist is an automated AI training system by Adaption Labs (founded by Sara Hooker, former VP of AI Research at Cohere) that automates the full research loop behind model training and alignment. The system co-optimizes training data and model recipes simultaneously until the model converges on a user-defined target behavior. According to internal benchmarks, AutoScientist outperforms human-configured training by an average of 35% and raised win rates from 48% to 64%. It targets ML engineers, non-technical builders, and enterprises seeking to reduce reliance on proprietary models or build their own specialized models.
Momentum-Verloop
04.04.03.07.
Features
| Deployment Model | Cloud-based platform (adaptionlabs.ai); per explainx.ai blog, on-premise deployment for regulated industries is also described (not confirmed on the official website) |
| Platform | Web platform (adaptionlabs.ai); benchmarks and fine-tuning runs were conducted using model architectures from Together AI |
| Price | Free for the first 30 days after launch; no public pricing published thereafter |
| Release Date | May 13, 2026 |