This paper introduces QINCo, a neural network-based residual vector quantizer that adaptively constructs specialized codebooks for each quantization step, improving the accuracy of vector compression and search. Unlike conventional residual quantization (RQ), which uses a fixed codebook per step, QINCo employs a neural network to generate codebooks that depend on the approximation of the vector from previous steps. This approach reduces the mean squared error (MSE) and enhances the overall performance. Experiments on multiple datasets and code sizes show that QINCo outperforms state-of-the-art methods, achieving better nearest-neighbor search accuracy with shorter codes. QINCo is also compatible with fast approximate search techniques, making it suitable for high-precision large-scale similarity search. The paper discusses the implementation details, including the use of inverted file indexes (IVF) and approximate decoding, and provides a comprehensive experimental evaluation, demonstrating the effectiveness of QINCo in various scenarios.This paper introduces QINCo, a neural network-based residual vector quantizer that adaptively constructs specialized codebooks for each quantization step, improving the accuracy of vector compression and search. Unlike conventional residual quantization (RQ), which uses a fixed codebook per step, QINCo employs a neural network to generate codebooks that depend on the approximation of the vector from previous steps. This approach reduces the mean squared error (MSE) and enhances the overall performance. Experiments on multiple datasets and code sizes show that QINCo outperforms state-of-the-art methods, achieving better nearest-neighbor search accuracy with shorter codes. QINCo is also compatible with fast approximate search techniques, making it suitable for high-precision large-scale similarity search. The paper discusses the implementation details, including the use of inverted file indexes (IVF) and approximate decoding, and provides a comprehensive experimental evaluation, demonstrating the effectiveness of QINCo in various scenarios.