Recently, IBM announced the launch of version 3.1 of its open source language model Granite, which brings many important enhancements. The new version of the model has been redesigned with a denser architecture, capable of handling up to 128,000 tokens at once. This update signifies a substantial improvement in Granite's ability to process complex text and tasks. The Granite 3.1 model has been trained on datasets in 12 languages and 116 programming languages, processing a total of 12 trillion tokens.

Recently at the Google I/O Berlin conference, Google announced the open-sourcing of its latest language model Gemma2, which has made significant breakthroughs in performance and efficiency. Gemma2 offers two parameter sizes: 9B and 27B, with the 27B version's performance approaching that of the 70B Llama3 model, but with a model size that is only about 40% of the latter.

Finnish startup Silo AI has released Poro, aiming to enhance multilingual AI capabilities in Europe. Poro is an open-source model that covers 24 official languages of the EU, with a parameter count of 3.42 billion. Poro uses cross-language training methods to address the performance challenges in training low-resource languages.

Startup Gradient has successfully raised $10 million to expand the adoption of its cloud platform. Developers can host and customize open source language models on this platform. Gradient aims to simplify AI infrastructure management by offering large language models and custom datasets. The platform is already being used by around 20 companies and plans to grow its team.