This research paper presents an integrated system for processing Arabic and English texts using Natural Language Processing (NLP) techniques and deep language models. The system is designed to correct grammatical and spelling errors, restore punctuation, remove unnecessary repetition, and preserve the original meaning of the text. It also includes a multilingual text classification module that categorizes texts into six major domains: medical, legal, academic, news, entertainment, and social. The study aimed to evaluate the proposed system from two main perspectives: linguistic correction quality and inference performance under two computing environments, Central Processing Unit (CPU) and Graphics Processing Unit (GPU with CUDA), while comparing sequential and parallel execution modes. Experiments were conducted on Arabic and English text datasets of different sizes, and output quality was measured using BLEU and GLEU metrics. The results showed that the proposed system outperformed the baseline approach, especially for Arabic texts, achieving average BLEU and GLEU scores of 0.7454 and 0.7529, respectively. For English texts, the system achieved 0.8143 BLEU and 0.8214 GLEU. In addition, parallel processing significantly reduced execution time in both CPU and GPU environments, with the best performance obtained using medium batch sizes between 64 and 128. The study also confirmed full consistency between sequential and parallel outputs, with a 100% matching rate. These findings indicate that the proposed system provides an effective solution that combines high correction quality with efficient execution speed, making it suitable for proofreading, text analytics, and large-scale text processing applications............... Keywords:............. Natural Language Processing, Text Correction, Text Classification, Parallel Processing, CPU, GPU, BLEU, GLEU.