diff --git a/convert_kytea_model/Cargo.toml b/convert_kytea_model/Cargo.toml
index d827d547..070740dd 100644
--- a/convert_kytea_model/Cargo.toml
+++ b/convert_kytea_model/Cargo.toml
@@ -6,4 +6,4 @@ edition = "2021"
 [dependencies]
 clap = { version = "4.2", features = ["derive"] }  # MIT or Apache-2.0
 vaporetto = { path = "../vaporetto", features = ["kytea"] }  # MIT or Apache-2.0
-zstd = "0.13"  # MIT
+zstd = { version = "0.13", features = ["zstdmt"] }  # MIT
diff --git a/convert_kytea_model/src/main.rs b/convert_kytea_model/src/main.rs
index b97efd25..b390e4af 100644
--- a/convert_kytea_model/src/main.rs
+++ b/convert_kytea_model/src/main.rs
@@ -19,6 +19,10 @@ struct Args {
     /// Vaporetto model file
     #[clap(long)]
     model_out: PathBuf,
+
+    /// The number of workers for zstd (0 means multithreaded will be disabled)
+    #[arg(long, default_value="0")]
+    zstd_workers: u32,
 }
 
 fn main() -> Result<(), Box<dyn std::error::Error>> {
@@ -31,6 +35,7 @@ fn main() -> Result<(), Box<dyn std::error::Error>> {
     eprintln!("Saving model file...");
     let model = Model::try_from(model)?;
     let mut f = zstd::Encoder::new(fs::File::create(args.model_out)?, 19)?;
+    f.multithread(args.zstd_workers)?;
     model.write(&mut f)?;
     f.finish()?;
 
diff --git a/manipulate_model/Cargo.toml b/manipulate_model/Cargo.toml
index a1c45b17..1e966a23 100644
--- a/manipulate_model/Cargo.toml
+++ b/manipulate_model/Cargo.toml
@@ -8,4 +8,4 @@ clap = { version = "4.2", features = ["derive"] }  # MIT or Apache-2.0
 csv = "1.2"  # Unlicense or MIT
 serde = { version = "1.0", features = ["derive"] }  # MIT or Apache-2.0
 vaporetto = { path = "../vaporetto" }  # MIT or Apache-2.0
-zstd = "0.13"  # MIT
+zstd = { version = "0.13", features = ["zstdmt"] }  # MIT
diff --git a/manipulate_model/src/main.rs b/manipulate_model/src/main.rs
index d5348452..35d9c6e7 100644
--- a/manipulate_model/src/main.rs
+++ b/manipulate_model/src/main.rs
@@ -23,6 +23,10 @@ struct Args {
     /// Replace a dictionary if the argument is specified.
     #[arg(long)]
     replace_dict: Option<PathBuf>,
+
+    /// The number of workers for zstd (0 means multithreaded will be disabled)
+    #[arg(long, default_value="0")]
+    zstd_workers: u32,
 }
 
 #[derive(Deserialize, Serialize)]
@@ -72,6 +76,7 @@ fn main() -> Result<(), Box<dyn std::error::Error>> {
     if let Some(path) = args.model_out {
         eprintln!("Saving model file...");
         let mut f = zstd::Encoder::new(fs::File::create(path)?, 19)?;
+        f.multithread(args.zstd_workers)?;
         model.write(&mut f)?;
         f.finish()?;
     }
diff --git a/train/Cargo.toml b/train/Cargo.toml
index e14f2561..fa145d5d 100644
--- a/train/Cargo.toml
+++ b/train/Cargo.toml
@@ -7,4 +7,4 @@ edition = "2021"
 clap = { version = "4.2", features = ["derive"] }  # MIT or Apache-2.0
 vaporetto = { path = "../vaporetto", features = ["train"] }  # MIT or Apache-2.0
 vaporetto_rules = { path = "../vaporetto_rules" }  # MIT or Apache-2.0
-zstd = "0.13"  # MIT
+zstd = { version = "0.13", features = ["zstdmt"] }  # MIT
diff --git a/train/src/main.rs b/train/src/main.rs
index 7958935a..94001908 100644
--- a/train/src/main.rs
+++ b/train/src/main.rs
@@ -65,6 +65,10 @@ struct Args {
     /// Do not normalize training data.
     #[arg(long)]
     no_norm: bool,
+
+    /// The number of workers for zstd (0 means multithreaded will be disabled)
+    #[arg(long, default_value="0")]
+    zstd_workers: u32,
 }
 
 fn main() -> Result<(), Box<dyn std::error::Error>> {
@@ -179,6 +183,7 @@ fn main() -> Result<(), Box<dyn std::error::Error>> {
     eprintln!("Finish training.");
 
     let mut f = zstd::Encoder::new(File::create(args.model)?, 19)?;
+    f.multithread(args.zstd_workers)?;
     model.write(&mut f)?;
     f.finish()?;