6 Minutes Read | Listen to Article
Elon Musk is opening the doors to his AI playground. xAI’s Grok 2.5, the company’s flagship language model until last year, is now open source, and Musk promises that Grok 3 will follow in about six months. Announced on X, formerly Twitter, he wrote, “The @xAI Grok 2.5 model, which was our best model last year, is now open source. Grok 3 will be made open source in about 6 months.” Grok 3, unveiled in February, currently holds the title of xAI’s most advanced model, while Musk hints that Grok 5 is on the horizon before the year ends.
Grok 2.5 is now available on Hugging Face, giving developers the freedom to download, run, and tweak it—though with a caveat: the code cannot be used to train competing AI models. This isn’t xAI’s first open-source venture; Grok 1 went public last March, offering researchers a playground to explore its raw capabilities. The move positions xAI in direct contrast with OpenAI, Musk’s former venture and now a key rival, which only recently released smaller open-source models, keeping its most powerful versions, like ChatGPT, under lock and key.
Grok hasn’t been without its hiccups. In the past, it generated problematic responses, including offensive remarks, which xAI attributed to outdated code and quickly corrected. Still, the incident highlighted the challenges of safely sharing powerful AI models.
By releasing Grok 2.5 to the public, Musk is inviting developers to innovate, troubleshoot, and expand on xAI’s work. The open-source path for Grok 3 and eventually Grok 5 will reveal whether xAI can consistently live up to its promise of transparency in an industry often shrouded in secrecy.
**This news was published on India Today on 25th August, 2025.
Source Click
Meta CEO Mark...