diff --git a/.github/scripts/strategy-matrix/generate.py b/.github/scripts/strategy-matrix/generate.py
index d2547018ce..8c8bcd24f3 100755
--- a/.github/scripts/strategy-matrix/generate.py
+++ b/.github/scripts/strategy-matrix/generate.py
@@ -179,8 +179,8 @@ def generate_strategy_matrix(all: bool, config: Config) -> list:
                 # Disable mold linker - it's too strict about relocations with GCC runtime libs
                 # Use default linker (bfd/ld) which is more lenient with mixed code models
                 cmake_args += ' -Duse_mold=OFF'
-                exe_linker_flags += ' -fuse-ld=bfd -static-libubsan -static-libasan -static-libtsan'
-                shared_linker_flags += ' -fuse-ld=bfd -static-libubsan -static-libasan -static-libtsan'
+                exe_linker_flags += ' -fuse-ld=gold'# -static-libubsan -static-libasan -static-libtsan'
+                shared_linker_flags += ' -fuse-ld=gold'# -static-libubsan -static-libasan -static-libtsan'
 
             if architecture['platform'] == 'linux/amd64' and os['compiler_name'] == 'gcc':
                 # Add -mcmodel=large and -fPIC to both compiler AND linker flags
@@ -218,8 +218,7 @@ def generate_strategy_matrix(all: bool, config: Config) -> list:
             # Also tsan doesn't work well mcmode=large and bfd linker
             if os['compiler_name'] == 'gcc':
                 extra_warning_flags += ' -Wno-tsan'
-                exe_linker_flags = exe_linker_flags.replace('-fuse-ld=bfd', '-fuse-ld=gold')
-                shared_linker_flags = shared_linker_flags.replace('-fuse-ld=bfd', '-fuse-ld=gold')
+                cxx_flags = cxx_flags.replace('-mcmodel=large', '-mcmodel=medium')
                 exe_linker_flags = exe_linker_flags.replace('-mcmodel=large', '-mcmodel=medium')
                 shared_linker_flags = shared_linker_flags.replace('-mcmodel=large', '-mcmodel=medium')