DeepSeek releases V3, a massive 671-billion parameter mixture-of-experts model with enhanced reasoning capabilities, mathematical problem-solving, and code generation at competitive costs.
Alibaba's latest Qwen 2.5-Max model delivers unprecedented performance in the open-source AI landscape, matching GPT-4 capabilities while offering superior efficiency, multilingual support, and complete transparency for developers and researchers worldwide.
Alibaba releases Qwen 2.5-Coder, a specialized 32-billion parameter open-source model designed for code generation, debugging, and software development with state-of-the-art programming capabilities.
Meta releases Llama 3.2 with groundbreaking vision capabilities and lightweight models optimized for edge deployment, bringing multimodal AI to mobile devices and IoT applications.