The grok-2-image-1212 AI model is now available via API and can be used to generate images natively.
The API revises the prompt using a chat model before sharing it with the image generation model
Photo Credit: X/Grok
xAI, the Elon Musk-led artificial intelligence (AI) firm behind Grok, released a new application programming interface (API) on Wednesday. The new API is unique as it is the first developer tool that supports image generation. The company has lately been focused on developers and has released as many as five APIs since the first one arrived in November 2024. The pricing is said to be slightly on the steeper end, and the company is currently not letting users customise the output.
Before the latest addition, xAI offered four customised AI models as API. Two were based on the first iteration of the Grok large language model (LLM) and two were based on Grok 2. While the company offered the image understanding feature, there was no way to generate images using the API.
This was likely because xAI was outsourcing the image generation capability available via its chat platform. Till last year, image generation on Grok was handled by the AI startup Black Forest Labs. However, in December, the AI firm announced the release of Aurora, an image generation model built on the mixture of experts (MoE) network. It appears the company is now expanding the model to developers as well.
In its documentation page, xAI now lists a new API model dubbed ‘grok-2-image-1212' which comes with image generation capability. The functionality is straightforward. Once a text prompt is sent, a chat model takes the instruction and revises the prompt for better clarity. The revised prompt is shared with the image generation model and the output is generated.
Currently, developers can generate up to 10 images with a single request by adjusting the parameter. There is a limit of five requests per second, crossing which will return an error message. The images are shared in JPEG format. A TechCrunch report claimed that xAI will charge $0.07 (roughly Rs. 6) per image.
The pricing places the service on the higher end, given that Black Forest Labs' Flux API costs $0.05 (roughly Rs. 4) per image and Google's Imagen 3 costs $0.03 (roughly Rs. 2.5) per image. However, Ideogram is still more expensive at $0.08 (roughly Rs. 7) per image.
Additionally, xAI highlighted that the API currently does not support output customisation such as changing the quality, size, or style of the images. Notably, the API's endpoint is compatible with OpenAI SDK and users can use the same base_url. However, it is not compatible with the Anthropic SDK.
Catch the latest from the Consumer Electronics Show on Gadgets 360, at our CES 2026 hub.