The current work investigates the formation of Nitric Oxide (NO) in hydrogen–air flames, over a wide range of flame temperatures. The use of hydrogen allows improved focus on the thermal-NO pathway by removing the complexity introduced by the prompt-NO pathway, which has been shown to be an important contributor to inaccurate predictions of absolute post-flame NO concentrations in hydrocarbon flames. This experimental study is conducted at atmospheric pressure using stoichiometric, premixed, laminar stagnation flames. Adiabatic flame temperatures ranging from 1600K to 2300K are achieved by varying the argon concentration in air. One-dimensional velocity, temperature, and NO concentration profiles are measured using non-intrusive laser diagnostics: Particle Tracking Velocimetry (PTV), NO multiline thermometry, and NO Laser Induced Fluorescence (NO-LIF), respectively. Results show that the experimental velocity profiles are incorrectly captured by the studied mechanisms, especially at low and high temperatures. This suggests that major inaccuracies are present in the hydrogen oxidation chemistry of the thermochemical models, regardless of their optimisation methodology. Furthermore, NO-LIF profiles show major discrepancies between all the studied mechanisms and the experiments, especially at elevated temperatures. The disagreement stems from an inaccurate description of the base chemistry of the models. These inaccuracies arise specifically from the description of the radical pool driving the flame behaviour and NO formation. This study demonstrates the need for model optimisation on experimental measurements using pure hydrogen 1D flames to obtain an accurate description of the hydrogen oxidation chemistry at play. This would lead to an improved description of the NOx sub-chemistry of any hydrogen, or hydrocarbon, combustion system.