Skip to main content
GPUBeat Section · 01

/Frontier Models

Releases. Benchmarks. Costs.

Stories
1,027
all-time
This week
1027
past 7d
Avg read
5.2m
rolling
OpenAI — ai-infrastructure — OpenAI, Anthropic
Frontier Models 4d

AI Agents Show Distinct Disparities in Browser Exploitation Capabilities

Carnegie Mellon researchers unveil a benchmark highlighting the stark performance gap between Anthropic's Claude Mythos and OpenAI's GPT-5.5 in exploiting browser vulnerabilities.

The archive

Page 68 · sorted by latest
Showing 1006–1020 of 1,027