Almost every developer announcement at Google I/O references AI, from text summarization to image recognition, reflecting the significant AI emphasis.
Google introduces Gemini 1.5 Pro and Gemini 1.5 Flash with one million token context windows, and plans for Gemma models expansion, including PaliGemma for multimodal vision-language tasks.
Gemini API improvements include support for parallel function calling and native video frame extraction, along with upcoming Context Caching to manage costs efficiently.
Google to launch Gemma 2 series in June with a first 27 billion parameter model optimized for NVIDIA GPUs or single TPU host, ensuring high performance.
Collection
[
|
...
]