Due to the deterioration and environmental issues of cement-based ultra-high-performance concrete (UHPC), there is an increasing interest in utilizing geopolymer technology to develop UHPC. In this regard, a statistically based model using response surface methods is applied to predict the fresh and mechanical properties of ultra-high performance geopolymer concrete (UHPGC) mixtures. Twenty mixtures were designed and generated experimentally using the central composite design (CCD) concept. In this context, the effects of three principal factors—the effects of silica fume (SF), natural sand content, and different curing temperatures—on the flowability, setting times, bulk density, compressive, and flexural strengths were examined. The models presented herein reveal that all the inputs and outputs are perfectly correlated. Microscopic investigations were used to determine the morphological characteristics of the optimum mixture. The results revealed that the bulk density, flexural, and compressive strengths of the UHPGC ranged from 2461 to 2536 kg/m3, 6.48–9.39 MPa, and 104–152 MPa, respectively. According to the validated models and optimization findings, the optimum mixture of SF (22.5%) and sand content (1150 kg/m3) cured at 100 °C was found to develop the optimized UHPGC with the best microstructure performance. From the experimental results, the optimized UHPGC mixture possessed excellent mechanical properties (2521 kg/m3, 9.39 MPa and 152 MPa), and it had a non-porous structure primarily made of a dense geopolymer gel. Overall, the current study provides new insights into the design and production of UHPGC materials using the response surface method, which could contribute to the widespread use of this concrete in practical applications.