To explore low-carbon spring wheat production strategies for the ecologically fragile Qinghai Plateau, a two-year field experiment was conducted using a split-plot design. The main plots were set with three nitrogen (N) application levels: N0 (0 kg·ha⁻¹, no N input), N1 (157.5 kg·ha⁻¹, 30% N reduction relative to conventional rate), and N2 (225 kg·ha⁻¹, the conventional rate). The subplots comprised four green manure mixed cropping patterns: sole cropped common vetch (CK), common vetch mixed with hull-less barley (HB), common vetch mixed with hairy vetch (HV), and common vetch mixed with rapeseed (RS). Averaged across green manure treatments, N1 reduced global warming potential (GWP) by 14.56% and greenhouse gas intensity (GHGI) by 16.89% compared with N2. Under the N1 treatment, compared with CK, RS reduced cumulative CO₂ emissions by 7.77% and GWP by 7.45%, maintained wheat grain yield comparable to those under N2, and obtained the minimum GHGI which was 2.92% lower than CK. Regarding soil properties, all green manure patterns significantly increased soil available phosphorus (AP) and available potassium (AK) compared with CK. Specifically, HB increased AP by 12.81% and AK by 16.38%, and RS also increased AP by 4.37% and AK by 2.36%. In addition, HB enhanced soil organic matter (SOM) by 5.85%, ammonium nitrogen (NH₄–N) by 8.13%, and nitrate nitrogen (NO₃–N) by 4.34%. Soil NH₄–N and NO₃–N were identified as key drivers of GHGI across treatments. On the Qinghai–Tibet Plateau, a 30% N reduction combined with target-oriented green manure mixtures—RS for emission mitigation and HB for high yield and fertility—offers a viable low-carbon strategy for spring wheat production.