Observed arrival · 2026-10-15
FrontierRSI’s 15-task autonomy benchmark
FrontierRSI says it evaluates AI agents on long-horizon machine-learning research tasks, using fixed budgets and independent tests.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- Researchers evaluating autonomous machine-learning agents
- Worth noticing
- The task list ranges from NMR prediction and no-backprop pretraining to Pokémon battles and watercolour painting.
Field notes
The homepage lists 15 task names spanning continual learning, representation control, NMR prediction, no-backprop pretraining, and even Pokémon battles and watercolour painting. It describes fixed budgets and independent tests, while the leaderboard is organized by model, overall score, cost, and a per-task Pareto view. In the captured page, named model rows are present but score and cost values are dashes, so comparative results are not visible.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
◎NicheUnusually specific use
One card from the complete issue