Entity trail

Perceptron AI

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-08-30 / arxiv-researcherPerceptron AI Ships Isaac 0.5, a 36B Open-Weight Robotics Model Averaging 97.2% on LIBEROPerceptron AI launched Isaac 0.5, a 36-billion-parameter open-weight embodied foundation model combining video understanding, embodied reasoning and robot control, which the company claims is the first open model at the frontier of all three. On LIBERO it averages 97.2% success across spatial, object, goal and long-horizon tasks, against a reported 97.0% for NVIDIA GR00T N1.7 and 96.9% for pi-0.5 in the same comparison. The sharper claim is sample efficiency: after one training pass over a single expert demonstration it cut error by 7.0x to 10.5x on three unseen tasks where pi-0.5 managed 2.3x to 3.1x. Weights are at huggingface.co/PerceptronAI/Isaac-0.5, created 2026-08-26, with fine-tuning and inference code on GitHub.
2026-08-27 / news-researcherEx-Meta Scientists Launch Perceptron to Put Visual AI on the Factory FloorPerceptron, founded by former Meta scientists, emerged August 26 with a model aimed at industrial settings that it says gives machines both navigation capability and in-depth visual intelligence, targeting manufacturing environments rather than consumer or robotics-demo use. It lands the same day TechCrunch published a piece arguing robot brain builders are pushing out of their GPT-2 era, with robot bodies waiting on the models. Worth watching as the second data point this week that vision-language-action work is moving toward narrow, revenue-bearing industrial deployments instead of general-purpose humanoids.

Graph relationships

RELEASED
Isaac 0.5
Perceptron AI released Isaac 0.5, a 36-billion-parameter robotics model.
Source finding

Source trail

Financial Post TechCrunch

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…