For calibration of flattening filter-free (FFF) linac beams, one needs to either use a chamber small enough to avoid volume averaging that occurs with the conventional Farmer-type chamber, or apply a gradient correction (Prp) to the Farmer chamber. Continued use of the Farmer chamber is attractive due to the desirable characteristics of the chamber. We measured Prp using two methods, the recommended profile gradient method, and an output method with a small chamber having negligible volume averaging, for both a Varian TrueBeam and Elekta Versa 6MV and 10MV FFF beams for 90 cm source-to-surface distance (SSD), 100 cm source-to-axis (SAD) calibrations. For profile-determined Prp, profiles were measured along the calibration plane (100 cm SAD, 90 cm SSD, 10 cm depth) using a diode having a sensitive diameter of 0.8mm. The profiles were integrated numerically along both the length and width of both chamber types, with the deviation from unity giving the magnitude of Prp. The Farmer chamber was 23 mm in length, 3.05 mm in radius, and had a volume of 0.6cc. The smaller chamber was 6.5 mm in length, 2.75 mm in radius, and a volume of 0.125cc. For the output-determined Prp, beam outputs were measured for 6MV conventional flattening filter (cFF), 6FFF, 10cFF, and 10FFF beams using the TG51 protocol for six Farmer and six 0.125cc chambers. Use of six chambers was thought to improve statistics. The 0.125cc chamber was not reference grade, and required cross calibration with the Farmer chamber using cFF beams. To validate the lack of a gradient for the cFF beams, profiles of the cFF beams were measured as well, with a corresponding profile-determined Prp. A ratio of the FFF beam output between Farmer and 0.125cc chambers gives the output-determined Prp. Values of profile-determined Prp for the TrueBeam were 1.003 and 1.007 for 6FFF and 10FFF beams, respectively. Values of profile-determined Prp for the Versa were 1.005 and 1.007 for 6FFF and 10FFF beams, respectively. Surprisingly, the Versa 10cFF beam used was not “flat”, having a central axis divot, giving a Prp 0.995. This beam was not included in the cross-calibration of the 0.125cc chambers. Values of the output-determined Prp were within 0.1% of the profile-determined Prp for the TrueBeam, and within 0.4% for the Versa. Repeated Versa measurements gave similar results. One reference for the 0.125cc chamber used here indicates an anomalous dose-per-pulse issue for cFF beams, which may be exacerbated for FFF beams. Values of Prp from other linacs available to us and from the literature indicate good agreement (within 0.1%, one 0.2% outlier) for the TrueBeam and profile-determined Prp for the Versa. Profile- and output-determined Prp values for the TrueBeam are consistent. A Prp correction for the Versa 10cFF may be warranted, but appears linac or energy dependent. The profile-determined Prp appears to be more reproducible.