Image via Unknown
Wednesday, December 3, 2025
OpenAI's Code Red: Meet Garlic, the Gemini Slayer
The competition is heating up in ways we haven't seen before. OpenAI just declared "code red" and fast-tracked a new model called Garlic to counter Google's Gemini surge (yikes), while Amazon dropped an absolute arsenal at re:Invent with new frontier agents, custom chips, and production platforms that could reshape how enterprises build. Meanwhile, Anthropic's Claude Code is scaling up with their Bun acquisition targeting that $1B revenue milestone, OpenAI revealed GPT-5.1's nuanced approach to balancing IQ and EQ with customizable tone, and Perplexity went full defensive mode releasing open-source tools to protect AI agents from sneaky prompt injection attacks lurking across the web (bold move on the security front). If Google's making OpenAI sweat this much, what's your next move?
Image via Unknown
Top Stories
OpenAI's emergency response to mounting competitive pressure from Google signals a pivotal shift in the AI race, with the company redirecting resources away from new products to fortify ChatGPT's core capabilities.
Anthropic
Anthropic acquires Bun to power Claude Code's explosive growth, which hit $1B run-rate revenue in just six months by combining AI-powered coding with optimized JavaScript infrastructure.
OpenAI
OpenAI discusses GPT-5.1's focus on combining reasoning capabilities with emotional intelligence and user-customizable model behavior, highlighting expanded tools for tuning model tone and personality while maintaining steerability and reducing bias.
AWS launched a complete AI infrastructure ecosystem including custom chips, frontier autonomous agents, expanded model families, and production-ready agent platforms, demonstrating 2-3x efficiency gains for enterprise customers across software development, customer service, and financial services workloads.
Perplexity
Perplexity open-sources BrowseSafe, a specialized prompt injection detection model and 14,000+ example benchmark to secure AI agents browsing the web, addressing emerging cybersecurity threats in agent-based browsing.
Keep Reading
Enjoyed this issue?
Get daily AI intel delivered to your inbox. No fluff, just the stories that matter.