Author: Gleisson Cabral

ChatGPT search is now accessible to everyone without a sign-up requirement, enhancing its usability across the globe.

OpenAI’s Deep Research feature is now fully rolled out to all Pro users across the UK, EU, Norway, Iceland, Liechtenstein, and Switzerland.

OpenAI has introduced new WhatsApp updates for ChatGPT, allowing users to upload images, use voice messages, and soon link their ChatGPT accounts for enhanced usage.

Google DeepMind has released details on safety measures for Gemini, including new reinforcement learning methods and red teaming to assess security risks.

Gemini 2.0 Flash-Lite is introduced, offering better quality than its predecessor at a similar cost and speed, with a 1 million token context window and multimodal input.

2.0 Pro Experimental is now available, excelling in coding and complex prompts, with a 2 million token context window.

2.0 Flash is optimized for high-volume tasks, showing improved performance and now available across more Google AI products.

Anthropic has increased the challenge for jailbreaking their system, offering $10K for passing all eight levels and $20K for a universal jailbreak.

@emollick notes that three models—Claude 3.5 Sonnet, o1-pro, and Gemini 2.0 Flash Thinking—have passed the Lem Test on the first try, indicating significant advancements in AI reasoning capabilities.

@MatthewBerman highlights the launch of Google’s Gemini 2.0 Flash Thinking Experimental, noting its impressive performance and free availability.

@bindureddy discusses the launch of Gemini 2.0, expressing optimism about its potential to outperform other models in performance metrics.