xAI Grok 2 large model (beta) released for X users

Last week, xAI announced Grok 2 and Grok 2 mini models and it’s now rolling out beta access for the large model on X social media site.

Following the announcement, the company integrated Grok 2 mini to X, allowing users to test the latest capabilities. However, the company decided to wait for the large model release. Now, you can choose between mini and large models with a premium subscription.

Grok is a large language model (LLM) launched in November 2023. The initial feature enabled users to conduct text-to-text conversations. Earlier this year, the company added a 1.5 update for the model with increased context length and improved performance.

The company has been building a new supercomputer to train Grok 2 models, which helped the company to improve this model and add new features.

xAI Grok 2 Beta (Source – Grok/X)

Grok 2 lets you generate images based on the text prompt. You need to describe the image in the prompt box and the model will give you an output within a few seconds.

You can share, regenerate, or download the image on your device to forward it to other platforms.

The new Grok version introduced it under the name sus-column-r into the LMSYS chatbot arena. It is a popular competitive language model benchmark. The company has seen immense improvement in performance compared to Claude and GPT-4 models on the leaderboard.

The added capabilities allow you to stay updated with the latest information. For example, you can get better results when you ask about a topic such as “What’s the latest news about Starship”.

The chatbot will gather sources from X and share an answer with reference. Grok has also become coding efficient for programmers, you can ask it to solve code problems and bugs in your code.

This is an early release for both models and the company has a vision update pending for the upcoming versions. That will allow you to attach an image, screenshot, or diagram to get an explanation, transcription, or learn about a specific part of the uploaded file.