MistralAI has launched its latest model, Mistral Small 3.1, featuring a 128,000 token context window and enhanced text performance, outperforming competitors such as Gemma 3 and GPT-4o Mini. The model operates efficiently on a Mac with just 32GB of RAM and achieves a processing speed of 150 tokens per second. Mistral Small 3.1 is fully open-source under the Apache 2.0 license. Meanwhile, Meta's Llama AI model has surpassed 1 billion downloads, indicating its widespread adoption among companies including Spotify and DoorDash. Meta is also preparing to unveil Llama 4, promising further advancements in AI capabilities.