Elon Musk’s xAI Releases Grok 3 in API for Builders, Introduces a ‘Fast’ Mannequin

Elon Musk’s xAI, the substitute intelligence (AI) agency behind the Grok chatbot, launched a brand new utility programming interface (API) on Thursday. The new API is for the Grok 3 AI mannequin, which the corporate launched in February. So far, the AI mannequin has been accessible to the top client through the X (previously often known as Twitter) platform, the devoted web site, or the cell apps. However, builders didn’t have entry to the newest fashions. With this, builders will have the ability to construct purposes and software program utilizing the Grok 3 AI mannequin.

xAI Releases Grok 3 API, Reveals Pricing

On its paperwork web page, xAI detailed the brand new API, which shall be powered by the Grok 3 AI fashions. The web page comprises mannequin overviews, pricing particulars, and extra info that shall be essential for builders who entry these fashions. In complete, there are 4 massive language fashions (LLMs) that builders will have the ability to entry through the brand new API — Grok 3 and Grok 3 mini (each accessible in beta), in addition to two sooner variants of the identical fashions.

The Grok 3 mannequin is the corporate’s flagship AI mannequin that can also be powering the chatbot on X, in addition to the apps. xAI mentioned that this mannequin will be useful in enterprise use instances akin to information extraction, coding, and textual content summarisation. On the opposite hand, the Grok 3 mini is a light-weight mannequin with native reasoning functionality. The AI agency claims it’s suited to logic-based duties that “do not require deep domain knowledge.”

Additionally, Grok 3 Fast beta and Grok 3 mini Fast beta are additionally accessible with the API. xAI mentioned each of those use the identical underlying mannequin because the non-fast variant, and ship equivalent response high quality. The important distinction is that the sooner variants are served on particular, sooner infrastructure and provide considerably sooner response instances in comparison with the usual mannequin. As a outcome, these fashions even have a better enter and output pricing.

Coming to pricing particulars, the Grok 3 mannequin will price builders $3 (roughly Rs. 260) per million enter tokens and $15 (roughly Rs. 1,300) per million output tokens. On the opposite hand, the sooner variant comes on the value of $5 (roughly Rs. 430) per million enter and $25 (roughly Rs. 2,150) per million output tokens.

The mini fashions are significantly cheaper in comparison with their customary counterparts. Grok 3 mini is priced at $0.30 (roughly Rs. 25) per million enter tokens, and $0.50 (roughly Rs. 43) per million output tokens. The Grok 3 mini quick AI mannequin will price builders $0.60 (roughly Rs. 51) for each million enter tokens, and $4 (roughly Rs. 344) for each million output tokens.

Notably, these AI fashions are text-only and can’t generate pictures. However, the fashions do settle for pictures as enter. Additionally, the API model of Grok 3 shouldn’t be related to the Internet, and the fashions are reliant on their pre-training datasets.