Google releases Gemma 4 12B model for consumer laptops
Google's Gemma 4 12B model can run on laptops with 16GB RAM, offering performance close to its 26B counterpart.
ARTICLES BY ALEX LINDGREN
LATEST NEWS
-
-
Google Deepmind's Gemma 4 12B Brings Multimodal AI to Laptops with 16 GB RAM
Google Deepmind's Gemma 4 12B model runs locally on laptops with just 16 GB of RAM and nearly matches a twice-as-large 26B model across benchmarks.
-
Ideogram 4.0 Released as Open-Weight Model with Native 2K Resolution
Ideogram 4.0, an open-weight text-to-image model, ranks first among open models on DesignArena, outperforming Midjourney v8 in a benchmark test.
-
xAI Launches Grok-Imagine-Video-1.5 Preview
xAI released Grok-Imagine-Video-1.5-preview on June 3, 2026, offering image-to-video generation at up to 720p resolution.
-
Microsoft Unveils MAI-Thinking-1, Tops Google in Image Generation
Microsoft's MAI-Thinking-1 matches leading models on software engineering benchmarks, placing second in image generation behind GPT-Image-2.
-
OpenAI Expands Codex With Role-Specific Plugins
OpenAI has added 62 apps and 110 capabilities to Codex, including role-specific plugins for data analysis, sales, and product design.
-
NVIDIA Expands Alpamayo Open Platform for Autonomous Vehicles
NVIDIA announced a major update to its Alpamayo platform, which now includes new models and tools for autonomous vehicle development, with over 400,000 downloads to date.
-
Anthropic Expands Claude Mythos to 15 Countries
Anthropic announced it is scaling its Claude Mythos AI model to 150 new organizations across 15 countries, including NATO and ENISA.
-
Lucidml's Supercut Model Generates Playable Games from Images
Lucidml's Supercut model creates playable games from images using consumer GPUs, with no training data from initial images.
-
Hugging Face Releases Borealis, 5B Audio-Language Model for Russian and English
Hugging Face announced Borealis, a 5B parameter open-source audio-language model trained on Russian and English data, achieving 20.88% WER on Russian benchmarks.
-
Google Unveils Gemini 3.5 Flash and Omni at I/O 2026
Google announced Gemini 3.5 Flash, a model that outperforms Gemini 3.1 Pro on benchmarks like Terminal-Bench 2.1 with 76.2% accuracy, at I/O 2026 on May 20, 2026.
-
DeepMind Launches Project Genie with Street View Integration
DeepMind has integrated Google Street View imagery into Project Genie, enabling users to create and explore realistic virtual environments based on real-world locations. The feature is now available for U.S. locations.