XAI, just a few months after the output of the grok 4 model Gok 4 Fast introduced the new artificial intelligence system bearing its name. This model uses 40 %less resources compared to Gok 4 and saves up to 98 %of the same scale tasks.
The new model is to work in long texts and complex tasks 2 million context window offers. In this way, operations such as research, writing and solutions can be processed at once. In addition, Gok 4 fast combines both thinking and directly answered tasks in a single structure. In addition, the model determines which task is to be processed with system redirects. However, this combined structure reduces the delays seen in previous versions. In this way, both faster response can be obtained and less transaction costs emerge.
Performance tests also reveal the capacity of Gok 4 Fast. 92 %in AIME 2025 examIn the HMMT test, over 93 %success has been achieved. In addition, the model gives nearly 4 Gok 4 results in different tasks. Especially in search -based comparisons, it is seen that it is at the top. These results show that Gok 4 fast maintains the same performance level as lower resource consumption. In addition, the model is in the top ten rankings on the text processing side.
GROK 4 FAST architecture increases productivity by maintaining performance
The XAI was not only focused on speed and efficiency when developing Gok 4 Fast. Security filters, content inspections and additional control mechanisms were added to the structure of the model. In the training process, large -scale data sets and user inputs were used. In addition, additional verification stages have been developed to prevent possible misunderstandings. The model was also strengthened by human feedback. Thus, measures were taken to maintain quality in content production.
There are also innovations in terms of user access. Gok 4 Fast, Web, iOS and Android, including free users can be used through. Free users can try “fast” and “automatic” modes for paid users. In particular, automatic mode reduces the response time by switching to the difficulty level of the problem. However, the XAI aims to provide similar experience to both groups of users. Thus, a wider section can benefit from the new model.
There is an important update on the cost side. The price determined for 1 million inputs is low. Exit costs are planned in parallel. A different pricing is applied in longer inputs. In addition, the cost is further with the cost of the inputs. This increases the usability of the model not only in performance but also in terms of budget.
Despite everything, the competition in the field of large language models does not slow down. Google’s Gemini series and Anthropic’s Claude Work on the models continues. For this reason, new comparisons will be made in the coming period to see the permanent impact of Gok 4 Fast in the market. For now, Gok 4 Fast stands out as one of the important moves of XAI with low cost and wide access opportunities.