The extent to which large language models (LLMs) can perform culturally grounded reasoning across non-English languages remains underexplored. This paper examines the reasoning and self-assessment abilities of LLMs across seven major Indian languages- Bengali, Gujarati, Hindi, Kannada, Malayalam, Tamil, and Telugu. We introduce a multilingual riddle dataset combining traditional riddles with context-reconstructed variants and evaluate five LLMs- Gemini 2.5 Pro, Gemini 2.5 Flash, Mistral-Saba, LLaMA-4-Scout, and LLaMA-4-Maverick under seven prompting strategies. In the first stage, we assess riddle-solving performance and find that while Gemini 2.5 Pro performs best overall, few-shot methods yield only marginal gains, and accuracy varies notably across languages. In the second stage, we conduct a self-evaluation experiment to measure reasoning consistency. The results reveal a key finding: a model’s initial accuracy is inversely correlated with its ability to identify its own mistakes. Top-performing models such as Gemini 2.5 Pro are overconfident (4.34% True Negative Rate), whereas lower-performing models like LLaMA-4-Scout are substantially more self-aware (42.09% True Negative Rate). These results point to clear gaps in multilingual reasoning and highlight the need for models that not only reason effectively but also recognize their own limitations.

Generating paintings using AI has been an intriguing area of research and has posed significant challenges in recent years. Landscape painting is a type of man-made ecological art form which contributes to preserving the ecological integrity of the environment we live in. Generative AI based Painting constitutes a form of visual expression encompassing various elements like drawings, arrangement, and conceptualization. Existing generative models do not replicate the painting process followed by a human painter. A human artist creates artwork in various stages such as: Sketching, Outlining and Colouring. Current generative models frequently restrict the range and diversity of styles by depending solely on carefully selected datasets such as WikiArt and VanGogh. The proposed work intends to utilize scraping techniques to collect a wide range of comprehensive and diverse landscape paintings. The primary objective of this research is to apply various generative AI models to generate artwork that replicate a human painting process and encompasses various artistic themes and styles instead of relying on a particular one. Performance of our work has shown that the landscape painting generation into distinct sketch and color phases have proven to be effective, fun and realistic.