xAI, the Elon Musk-led synthetic intelligence (AI) agency behind Grok, launched a brand new utility programming interface (API) on Wednesday. The brand new API is exclusive as it’s the first developer device that helps picture technology. The corporate has currently been centered on builders and has launched as many as 5 APIs because the first one arrived in November 2024. The pricing is claimed to be barely on the steeper finish, and the corporate is presently not letting customers customise the output.
xAI Introduces New API With Picture Technology Functionality
Earlier than the newest addition, xAI provided 4 customised AI fashions as API. Two have been primarily based on the primary iteration of the Grok massive language mannequin (LLM) and two have been primarily based on Grok 2. Whereas the corporate provided the picture understanding characteristic, there was no option to generate photographs utilizing the API.
This was doubtless as a result of xAI was outsourcing the picture technology functionality accessible by way of its chat platform. Until final yr, picture technology on Grok was dealt with by the AI startup Black Forest Labs. Nonetheless, in December, the AI agency introduced the discharge of Aurora, a picture technology mannequin constructed on the combination of specialists (MoE) community. It seems the corporate is now increasing the mannequin to builders as effectively.
In its documentation web page, xAI now lists a brand new API mannequin dubbed ‘grok-2-image-1212′ which comes with picture technology functionality. The performance is easy. As soon as a textual content immediate is distributed, a chat mannequin takes the instruction and revises the immediate for higher readability. The revised immediate is shared with the picture technology mannequin and the output is generated.
Presently, builders can generate as much as 10 photographs with a single request by adjusting the parameter. There’s a restrict of 5 requests per second, crossing which can return an error message. The pictures are shared in JPEG format. A TechCrunch report claimed that xAI will cost $0.07 (roughly Rs. 6) per picture.
The pricing locations the service on the upper finish, provided that Black Forest Labs’ Flux API prices $0.05 (roughly Rs. 4) per picture and Google’s Imagen 3 prices $0.03 (roughly Rs. 2.5) per picture. Nonetheless, Ideogram continues to be dearer at $0.08 (roughly Rs. 7) per picture.
Moreover, xAI highlighted that the API presently doesn’t help output customisation equivalent to altering the standard, dimension, or fashion of the photographs. Notably, the API’s endpoint is suitable with OpenAI SDK and customers can use the identical base_url. Nonetheless, it isn’t suitable with the Anthropic SDK.
