Skip to main content
Measured savings across 11 LLMs, from Claude Opus 4.7 to Gemini Flash.→ See per-model data
Connect your client
Research

Berkeley researchers tackle fragile long-horizon planning in learned world model

GRASP, a new gradient-based planner from UC Berkeley, addresses brittleness in long-horizon planning with learned world models by parallelizing optimization across time and reshaping gradients for cleaner action signals.

1 min read

Researchers at UC Berkeley have released GRASP, a gradient-based planning method designed to make long-horizon planning with learned world models practical. The work, conducted with collaborators including Yann LeCun and Mike Rabbat, tackles a fundamental problem in robotics and embodied AI: existin...

Sign in to read the full analysis

Free account. Full analysis on LLM unit economics, plus the weekly Cost-of-Inference column.

Method & sources
Source type
Primary publication (lab/vendor blog) — our analysis + implication
Source link
Berkeley AI Research
Published
UTC
Byline
By the gotcontext.ai team (editorial standards)
Correction?
corrections@gotcontext.ai