GEEK HAUS
← back to sources

Stories from blog.google

7 articles

·blog.google

Google DeepMind publishes model card for Gemini 3.8 Flash and a cyber-focused Gemini 3.8 Flash variant

Google DeepMind appears to have posted documentation for Gemini 3.8 Flash and Gemini 3.8 Flash Cyber via its model card site. The linked page likely outlines the models’ intended u...

read →
·blog.google

Google publishes Gemini API documentation for a new Gemini 3.7 Flash model

Google’s AI developer site now has documentation for Gemini 3.7 Flash, indicating a new Flash-tier Gemini model is available or being prepared for API use. The linked page appears...

read →
·blog.google

Google releases Gemma 4 QAT checkpoints to shrink on-device AI models for phones, laptops, and consumer GPUs

Google DeepMind released new Gemma 4 checkpoints optimized with Quantization-Aware Training to reduce memory needs while preserving model quality. The update includes support for Q...

read →
·blog.google

Google DeepMind unveils Gemma 4 12B, an encoder-free multimodal AI model built to run on laptops

Google DeepMind introduced Gemma 4 12B, a mid-sized multimodal model that routes vision and audio inputs directly into the LLM backbone without separate encoders. The model is desi...

read →
·blog.google

Google redesigns Search around Gemini, turning the search box into an AI chat and task assistant

Google is overhauling its iconic search box to make Gemini a more central part of Search, shifting more queries toward AI-generated answers and conversational follow-ups. Reports s...

read →
·blog.google

Google’s Gemini API documentation lists a new Gemini 3.5 Flash model for developers

Google’s AI developer documentation includes a page for Gemini 3.5 Flash, pointing to a new or upcoming model in the Gemini API lineup. The available source is the model documentat...

read →
·blog.google

Google releases multi-token prediction drafters for Gemma 4, promising up to 3x faster inference without quality loss

Google is adding Multi-Token Prediction drafters to the Gemma 4 open model family, using speculative decoding to reduce inference latency. The company says the approach can deliver...

read →