Grok 3 impresses with graphics but fails in coding

As expected, Elon Musk has significantly overestimated his new neural network. 

Grok 3, the new AI from xAI, surprised users with the quality of its generated images, but its programming skills leave much to be desired. Examples of successful illustrations have already appeared online, yet code generation has been disappointing.

Breakthrough in graphics

Users noted that the new version of Grok 3 has significantly improved image generation. For instance, frames in 8-bit style, created upon request, look quite convincing and even detailed. This proves that the model is capable of correctly interpreting visual requests and delivering a decent result.

Weak link – programming

The situation is quite different with code. Developers who tested Grok 3's programming capabilities encountered errors and incorrect solutions. This is particularly noticeable when generating complex algorithms.

Some users suggest that the problem may be related to vague prompts, but for now, Grok 3 clearly lags behind its competitors in this area.

Prospects and refinements

Perhaps in future updates, xAI developers will be able to improve the quality of code generation, but for now, the model demonstrates uneven development. While artists and designers find uses for Grok 3, programmers continue to wait for improvements.