Gemini 2.5 Flash: 24k Context, Multimodal, Function Calls, $0.15/1M Tokens

New AI Model Releases & Updates

• Gemini 2.5 Flash Launch: Google DeepMind announced their new hybrid reasoning model with controllable “thinking” capabilities. The model features 24k token context window, multimodal input, and function calling capabilities. Pricing starts at $0.15 per 1M input tokens.

• OpenAI’s o3 and o4-mini Models: NVIDIA AI highlighted OpenAI’s release of new models with state-of-the-art performance in coding, math, and visual perception. Notably, o4-mini focuses on fast, efficient reasoning at scale.

Product & Platform Updates

• Google AI Studio Enhancements: Developers can now build, save and share their own apps directly in AI Studio, with 10+ starter apps, Google Drive integration, and automatic API key management.

• ChatGPT Improvements: Sam Altman noted significant improvements in ChatGPT’s capabilities, suggesting a comparison with the December 2022 version would show dramatic progress.

• LlamaIndex Integration: The platform now supports Gemini 2.5 flash without requiring any upgrades, enabling simple integration through model name calls.

Product Management Insights

• OKR Implementation: A thread discussed common OKR failures in product organizations, emphasizing their role in thinking rather than just reporting.

• Product Experimentation: Insights shared about turning assumptions into small, fast experiments based on 100+ product experiments (80+ failures).

• Leadership Through Influence: Discussion about developing the ability to lead through influence rather than authority as a critical PM skill.

Memes & Humor

• Product manager reaction after reading all the product management books & taking courses
• Accountability post about product launch schedules