The launch of EuroLLM-9B marks a significant advancement in multilingual AI models, supporting 35 languages, including all 24 official EU languages. This model has been trained on 4 trillion tokens of high-quality multilingual data and is noted for outperforming existing European LLMs, such as Gemma and Mistral. Additionally, Mistral's Pixtral Large has achieved recognition as the top open model in the Vision Arena, surpassing the Qwen-VL-72B model. The Pixtral Large features a 123 billion parameter multimodal decoder and a 1 billion parameter visual encoder, making it suitable for processing high-resolution images. Other recent developments in AI include the introduction of several new models, such as the bggpt-gemma-2-2.6b-it-v1.0, a Bulgarian language model, and Chatty-Harry_v3.0, available on LocalAI, showcasing the ongoing innovation in the AI landscape.
Pixtral models are performing well on LMsys: - Pixtral 12B is on par with Llama 3.2 90B - Pixtral Large 123B is the best open-weight vision model by a large margin https://t.co/ffVfOZIqxu
🎉 New model alert! 🎉 Chatty-Harry_v3.0 is now available on LocalAI! 🚀 Get it running with `local-ai run chatty-harry_v3.0`. 🤖 #LocalAI #ChattyHarry #NewModelComing 🚀
🚀 New model alert! Introducing "cybercore-qwen-2.1-7b"! Merged using TIES merge method with rombodawg/Rombos-LLM-V2.5-Qwen-7b as base. Install it in LocalAI with `local-ai run cybercore-qwen-2.1-7b`. #LocalAI #newmodel #AI 🤖