diff --git a/benchmarks/advection/README.md b/benchmarks/advection/README.md
index 3d2233891d7..37aee45dded 100644
--- a/benchmarks/advection/README.md
+++ b/benchmarks/advection/README.md
@@ -13,4 +13,4 @@ Benchmarks:
 2. Rotate Shape (rotate_shape_supg.prm, rotate_shape_ev.prm)
    Various shapes being rotated by 360 degrees by a fixed velocity
 
-See [doc/advection.md](doc/advection.md) for details.
+See [](doc/advection) for details.
diff --git a/benchmarks/advection/doc/advection.md b/benchmarks/advection/doc/advection.md
index 457796a9064..fff404b9705 100644
--- a/benchmarks/advection/doc/advection.md
+++ b/benchmarks/advection/doc/advection.md
@@ -23,7 +23,7 @@ Both benchmarks have the identical setup in the temperature and a
 compositional field. The only difference is that the temperature equation
 contains a (small) physical diffusion term.
 
-**[Description of benchmark files](../README.md)**
+**[Description of benchmark files](../README)**
 
 ```{figure-md} fig:benchmark-drop
 <img src="drop.png" />
diff --git a/benchmarks/rayleigh_taylor_instability/README.md b/benchmarks/rayleigh_taylor_instability/README.md
index 72b2d434f5a..f2802511331 100644
--- a/benchmarks/rayleigh_taylor_instability/README.md
+++ b/benchmarks/rayleigh_taylor_instability/README.md
@@ -11,4 +11,4 @@ This file can later be processed with the 'gnuplot_script'.
 There is also the standalone input file `rayleigh_taylor_instability.prm` which
 runs only one of the cases (isoviscous and lambda=256km).
 
-See [doc/rayleigh_taylor_instability.md](doc/rayleigh_taylor_instability.md) for more information.
+See [](doc/rayleigh_taylor_instability) for more information.
diff --git a/benchmarks/rayleigh_taylor_instability/doc/rayleigh_taylor_instability.md b/benchmarks/rayleigh_taylor_instability/doc/rayleigh_taylor_instability.md
index 8afd0970dc1..a7a8c6a51f6 100644
--- a/benchmarks/rayleigh_taylor_instability/doc/rayleigh_taylor_instability.md
+++ b/benchmarks/rayleigh_taylor_instability/doc/rayleigh_taylor_instability.md
@@ -57,7 +57,7 @@ file:
     end
 
 
-**[Description of benchmark files](../README.md)**
+**[Description of benchmark files](../README)**
 
 ```{figure-md} fig:RTi_grids_a
 <img src="grid.*" style="width:44.0%" />
diff --git a/benchmarks/sinking_block/README.md b/benchmarks/sinking_block/README.md
index e21d5ca4082..c76a8733794 100644
--- a/benchmarks/sinking_block/README.md
+++ b/benchmarks/sinking_block/README.md
@@ -8,4 +8,4 @@ In order to run the benchmark for many resolution, density and viscosity combina
 all the input files generated inside the triple loop.
 The `sinking_block.prm` input file is the standalone prm file for the sinking block benchmark.
 
-For more information see [doc/sinking_block.md](doc/sinking_block.md).
+For more information see [](doc/sinking_block).
diff --git a/benchmarks/sinking_block/doc/sinking_block.md b/benchmarks/sinking_block/doc/sinking_block.md
index ed7b247a911..ae5f76153e6 100644
--- a/benchmarks/sinking_block/doc/sinking_block.md
+++ b/benchmarks/sinking_block/doc/sinking_block.md
@@ -34,7 +34,7 @@ $\delta\rho=8,32,128~\text{ kg m}^{-3}$. Results are shown in
 {numref}`fig:sinking_block2` and we indeed recover the expected trend with all data
 points forming a single smooth line.
 
-**[Description of benchmark files](../README.md)**
+**[Description of benchmark files](../README)**
 
 ```{figure-md} fig:sinking_block1
 <img src="dens_vel.png" style="width:60.0%" />