Global atmospheric methane growth rate has increased dramatically since 2007, peaking in 2021, yet the relative roles of emissions and atmospheric oxidation remain uncertain. Accurate attribution is essential for climate mitigation because emissions-driven and chemistry-driven changes imply fundamentally different policy responses. Here we investigate global methane production and loss from 2018 to 2024 using an ensemble Kalman filter coupled to the GEOS‑Chem model. We employ two methane inversion configurations: assuming climatological monthly OH distributions or jointly optimizing zonal mean OH with methane. The joint inversion reveals substantial interannual variability in OH, including an ~18% decline in 2020 followed by a recovery in subsequent years. Accounting for this variability reduces the inferred 2019–2020 emission increase by ~63% (14±6 Tg/yr versus 37±5.5 Tg/yr with fixed OH), demonstrating that changes in OH strongly influences source attribution. The total increased methane loss, 2024 minus 2019, is about 31±6 Tg/yr, including a temperature‑driven increase of the OH+methane reaction rate that represents about 40% of the total sink increase. Most of the sink variations originate from the tropics, where the largest shifts in emission occur.  Despite year-to-year variations, emissions remain the primary driver of changes in the methane growth rate change, except in 2020.  Both inversions identify significant emission reductions over tropical South America in 2023–2024, likely linked to regional drought. Broadly, our results underscore the necessity of jointly estimating methane sources and sinks to interpret recent atmospheric methane trends.