Ai2 ToyBlocks evaluates robot learning with controlled block tasks
Ai2 released ToyBlocks, a controlled block-manipulation system that combines a vision-language high-level planner with a motion policy.
1 min readRead story
CHOOSE YOUR FOCUS
What do you create?
NEWS FEED
Ai2
TopicAi2Clear topic
Sorted by AI Radar publication time · Times shown in Istanbul (UTC+3).Clear filters
Ai2’s StateBench shows where long-context state tracking breaks down
StateBench evaluates evolving entity state across four difficulty levels; frontier models fall to 13% accuracy or lower on the longest, hardest cases.
2 min readRead story
Ai2 NoveltyBench stress-tests LLM judgments of research novelty
Ai2’s new benchmark shows that six LLM judges can reach sharply different conclusions about the novelty of scientific ideas after small changes to prompts and evaluation setup.
2 min readRead story
Ai2’s LitPivot reshapes research ideas together with the literature
LitPivot splits a research proposal into problem, contribution and evaluation facets, dynamically retrieves relevant work, and offers critiques and revision suggestions backed by source excerpts.
2 min readRead story
Ai2 releases open satellite embeddings for the whole planet with OlmoEarth v1.3
Ai2’s OlmoEarth v1.3 provides global satellite embeddings at 10-meter resolution and a browser-based map explorer alongside open weights, data and code.
2 min readRead story
BEYOND THE HEADLINE
What changed?
What can it bring to your work?
Every story includes a concise summary, its creative workflow implications and original sources. Explore official demos and examples alongside the news.
Our editorial approach
