Grok 4 is currently free for all users

Grok users, paying or not, have unlimited access to the new Grok 4 model for a limited time, xAI announced on Aug. 10 — a move to remain competitive as its fellow AI companies launch increasingly advanced models and generative AI tools.

“Simply use Auto mode, and Grok will route complex queries to Grok 4. Prefer control? Choose ‘Expert’ anytime to always use Grok 4,” xAI wrote in an announcement posted on X.

This Tweet is currently unavailable. It might be loading or has been removed.

As xAI vies for a larger user base, the company may be trying to poach disgruntled ChatGPT users following the launch of OpenAI’s GPT-5. On platforms like Reddit, many ChatGPT users have complained that GPT-5 feels less humanlike and is a little more careful in the way it approaches its relationships with users. After facing a backlash, OpenAI CEO Sam Altman announced the company will re-release its previous model, GPT-4o, allowing other AI companies to step in with less-controlled alternatives.

Early tests of Grok 4, which launched on July 9, showed it’s an impressively advanced model, surpassing competitors like DeepSeek and Claude, but the model’s apparent lack of guardrails also suggests the new bot could pose safety concerns for users. A Mashable investigation showed that xAI’s new generative AI image tool, Grok Imagine, failed to meet basic safety standards, allowing users to create sexual deepfakes.

How to access Grok 4 for free

You can start using Grok 4 now in the Grok iOS and Android app, or via the X website.

While on auto mode, the chatbot will evaluate whether or not a prompt requires the extended responses of Grok 4 or the quicker feedback of Grok 3. Users can also select the “Expert” setting to get answers from Grok 4.

The company will also be expanding its usage limits for users to accommodate the new model. Further information on free tiers for Grok 4 has yet to be provided.

Disclosure: Ziff Davis, Mashable’s parent company, in April filed a lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.