AI Frontier Weekly Content Archive
8 posts · Page 1 of 1
- This Week in Frontier AI: Smaller Guardrails, Cheaper GPT-5.6, and a Sandbox Breach
- This Week in Frontier AI: GLM-5.3-Flash, FrontierCode, and Spoofed Logs
- OpenAI's agents hit a trust bottleneck
- A no-name model tied GPT-6 Astra, and Claude now leads 26% of Anthropic's AI R&D
- Claude Opus 5.5 tops the index, GPT-6 prices halve, and OpenAI pauses training
- Gemini 4 Argon tops a coding benchmark, and OpenAI shelves GPT-6.1 Astra
- Mistral previews a 1-trillion-parameter model, and Claude Haiku 5.5 costs 90% less
- Meta's first frontier model lands two points behind Claude Opus 5.5, and Gemini 4 Argon opens to subscribers
Explore more channels on Discover