Google Gemma 4
Modelos de IA de última geração, projetados para raciocínio avançado, oferecendo flexibilidade e controle total aos desenvolvedores.
Gemma 4 oferece modelos de IA abertos avançados, ideais para raciocínio e fluxos de trabalho agentivos. Disponível em vários tamanhos, com licença Apache 2.0.
Mais sobre Google Gemma 4
Google Gemma 4
Gemma 4 is a new family of open models designed for advanced reasoning and agentic workflows. These models provide a high level of intelligence-per-parameter, complementing Gemini models. They are available in various sizes and are suitable for diverse hardware, from mobile devices to high-performance workstations.
- •Advanced Reasoning Capabilities:Gemma 4 excels at multi-step planning and complex logic, showing significant improvements in math and instruction-following benchmarks.
- •Agentic Workflows Support:It natively supports function-calling, structured JSON output, and system instructions, enabling the building of autonomous agents for reliable workflow execution.
- •Code Generation Assistance:Supports high-quality offline code generation, transforming workstations into local AI code assistants.
- •Multimodal Processing:All models natively process video and images, supporting variable resolutions, and excel at visual tasks like OCR and chart understanding.
- •Long Context Windows:Edge models feature a 128K context window, while larger models offer up to 256K, allowing processing of extensive content in a single prompt.
- •Extensive Language Support:Trained on over 140 languages, enabling developers to build inclusive, high-performance applications for a global audience.
- •Mobile and IoT Optimization:E2B and E4B models are engineered for maximum compute and memory efficiency, running offline with near-zero latency on edge devices.
- •Open-Source License:Released under an Apache 2.0 license, providing complete developer flexibility and control over data, infrastructure, and models.
By choosing Gemma 4, you gain access to powerful, open-source models that transform how you approach AI development. These models offer advanced reasoning, agentic workflows, and multimodal capabilities, empowering you to create innovative applications. With its open-source nature, you have the freedom to build, deploy, and customize the models to meet your specific needs, driving positive change in the world.
