---
title: "OpenAI's LifeVal benchmark; Hydrogen-1.0 open science model"
date: 2026-11-02
updated: 2026-11-02T00:00:00Z
url: https://arun477.github.io/feed/2026-11-02/
author: Arun Arumugam
---

# OpenAI's LifeVal benchmark; Hydrogen-1.0 open science model

OpenAI measures everyday personal-agent tasks; Deep Principle opens a GLM-5.3-based science model.


- [OpenAI introduces LifeVal, a benchmark for everyday personal-assistant tasks](https://openai.com/index/lifeval/) (OpenAI, news): New eval of real-life tasks (childcare, groceries, government paperwork) done via browser, computer use and connectors. GPT-6 Astra scores 80.6% vs 49.3% for GPT-5.2. Self-reported; no task count, grader details or other labs.
- [Deep Principle releases Hydrogen-1.0, an open-weight science model post-trained from GLM-5.3](https://blog.deepprinciple.com/introducing-hydrogen-1.0/) (Deep Principle, release): MIT-licensed weights (1M context) for chemistry, materials and life sciences, trained on lab records and agent trajectories. Self-reported 38.3% on FrontierScience-Research vs GPT-6 Astra's 33.3%; not yet independently verified.