diff --git a/crates/ruff_linter/src/importer/insertion.rs b/crates/ruff_linter/src/importer/insertion.rs
index 5cd6ae200c21c..274147a756e84 100644
--- a/crates/ruff_linter/src/importer/insertion.rs
+++ b/crates/ruff_linter/src/importer/insertion.rs
@@ -321,7 +321,6 @@ mod tests {
 
     use ruff_python_ast::PySourceType;
     use ruff_python_codegen::Stylist;
-    use ruff_python_parser::lexer::LexResult;
     use ruff_python_parser::{parse_suite, Mode};
     use ruff_source_file::{LineEnding, Locator};
     use ruff_text_size::TextSize;
@@ -332,7 +331,7 @@ mod tests {
     fn start_of_file() -> Result<()> {
         fn insert(contents: &str) -> Result<Insertion> {
             let program = parse_suite(contents)?;
-            let tokens: Vec<LexResult> = ruff_python_parser::tokenize(contents, Mode::Module);
+            let tokens = ruff_python_parser::tokenize(contents, Mode::Module);
             let locator = Locator::new(contents);
             let stylist = Stylist::from_tokens(&tokens, &locator);
             Ok(Insertion::start_of_file(&program, &locator, &stylist))
@@ -443,7 +442,7 @@ x = 1
     #[test]
     fn start_of_block() {
         fn insert(contents: &str, offset: TextSize) -> Insertion {
-            let tokens: Vec<LexResult> = ruff_python_parser::tokenize(contents, Mode::Module);
+            let tokens = ruff_python_parser::tokenize(contents, Mode::Module);
             let locator = Locator::new(contents);
             let stylist = Stylist::from_tokens(&tokens, &locator);
             Insertion::start_of_block(offset, &locator, &stylist, PySourceType::default())
diff --git a/crates/ruff_linter/src/linter.rs b/crates/ruff_linter/src/linter.rs
index cb3ac3318b784..ab94efe3fd09a 100644
--- a/crates/ruff_linter/src/linter.rs
+++ b/crates/ruff_linter/src/linter.rs
@@ -4,7 +4,7 @@ use std::path::Path;
 
 use anyhow::{anyhow, Result};
 use colored::Colorize;
-use itertools::Itertools;
+use itertools::{Either, Itertools};
 use log::error;
 use rustc_hash::FxHashMap;
 
@@ -14,7 +14,7 @@ use ruff_python_ast::{PySourceType, Suite};
 use ruff_python_codegen::Stylist;
 use ruff_python_index::Indexer;
 use ruff_python_parser::lexer::LexResult;
-use ruff_python_parser::{AsMode, ParseError};
+use ruff_python_parser::{AsMode, ParseError, TokenKind, Tokens};
 use ruff_source_file::{Locator, SourceFileBuilder};
 use ruff_text_size::Ranged;
 
@@ -353,7 +353,7 @@ pub fn add_noqa_to_path(
     let contents = source_kind.source_code();
 
     // Tokenize once.
-    let tokens: Vec<LexResult> = ruff_python_parser::tokenize(contents, source_type.as_mode());
+    let tokens = ruff_python_parser::tokenize(contents, source_type.as_mode());
 
     // Map row and column locations to byte slices (lazily).
     let locator = Locator::new(contents);
@@ -518,8 +518,7 @@ pub fn lint_fix<'a>(
     // Continuously fix until the source code stabilizes.
     loop {
         // Tokenize once.
-        let tokens: Vec<LexResult> =
-            ruff_python_parser::tokenize(transformed.source_code(), source_type.as_mode());
+        let tokens = ruff_python_parser::tokenize(transformed.source_code(), source_type.as_mode());
 
         // Map row and column locations to byte slices (lazily).
         let locator = Locator::new(transformed.source_code());
@@ -715,7 +714,7 @@ impl<'a> ParseSource<'a> {
 #[derive(Debug, Clone)]
 pub enum TokenSource<'a> {
     /// Use the precomputed tokens to generate the AST.
-    Tokens(Vec<LexResult>),
+    Tokens(Tokens),
     /// Use the precomputed tokens and AST.
     Precomputed {
         tokens: &'a [LexResult],
@@ -723,6 +722,17 @@ pub enum TokenSource<'a> {
     },
 }
 
+impl TokenSource<'_> {
+    pub fn kinds(&self) -> impl Iterator<Item = TokenKind> + '_ {
+        match self {
+            TokenSource::Tokens(tokens) => Either::Left(tokens.kinds()),
+            TokenSource::Precomputed { tokens, .. } => {
+                Either::Right(tokens.iter().flatten().map(|(tok, _)| tok.kind()))
+            }
+        }
+    }
+}
+
 impl Deref for TokenSource<'_> {
     type Target = [LexResult];
 
@@ -743,7 +753,7 @@ impl<'a> TokenSource<'a> {
     ) -> Result<AstSource<'a>, ParseError> {
         match self {
             Self::Tokens(tokens) => Ok(AstSource::Ast(ruff_python_parser::parse_program_tokens(
-                tokens,
+                tokens.into_inner(),
                 source_kind.source_code(),
                 source_type.is_ipynb(),
             )?)),
diff --git a/crates/ruff_linter/src/rules/pyflakes/mod.rs b/crates/ruff_linter/src/rules/pyflakes/mod.rs
index 8ef30efbdc264..b4f361809878b 100644
--- a/crates/ruff_linter/src/rules/pyflakes/mod.rs
+++ b/crates/ruff_linter/src/rules/pyflakes/mod.rs
@@ -11,7 +11,6 @@ mod tests {
 
     use anyhow::Result;
     use regex::Regex;
-    use ruff_python_parser::lexer::LexResult;
 
     use test_case::test_case;
 
@@ -591,7 +590,7 @@ mod tests {
         let source_type = PySourceType::default();
         let source_kind = SourceKind::Python(contents.to_string());
         let settings = LinterSettings::for_rules(Linter::Pyflakes.rules());
-        let tokens: Vec<LexResult> = ruff_python_parser::tokenize(&contents, source_type.as_mode());
+        let tokens = ruff_python_parser::tokenize(&contents, source_type.as_mode());
         let locator = Locator::new(&contents);
         let stylist = Stylist::from_tokens(&tokens, &locator);
         let indexer = Indexer::from_tokens(&tokens, &locator);
diff --git a/crates/ruff_linter/src/test.rs b/crates/ruff_linter/src/test.rs
index d83c5f54030f7..63a69f3857a90 100644
--- a/crates/ruff_linter/src/test.rs
+++ b/crates/ruff_linter/src/test.rs
@@ -16,7 +16,6 @@ use ruff_notebook::NotebookError;
 use ruff_python_ast::PySourceType;
 use ruff_python_codegen::Stylist;
 use ruff_python_index::Indexer;
-use ruff_python_parser::lexer::LexResult;
 use ruff_python_parser::AsMode;
 use ruff_python_trivia::textwrap::dedent;
 use ruff_source_file::{Locator, SourceFileBuilder};
@@ -111,8 +110,7 @@ pub(crate) fn test_contents<'a>(
     settings: &LinterSettings,
 ) -> (Vec<Message>, Cow<'a, SourceKind>) {
     let source_type = PySourceType::from(path);
-    let tokens: Vec<LexResult> =
-        ruff_python_parser::tokenize(source_kind.source_code(), source_type.as_mode());
+    let tokens = ruff_python_parser::tokenize(source_kind.source_code(), source_type.as_mode());
     let locator = Locator::new(source_kind.source_code());
     let stylist = Stylist::from_tokens(&tokens, &locator);
     let indexer = Indexer::from_tokens(&tokens, &locator);
@@ -177,7 +175,7 @@ pub(crate) fn test_contents<'a>(
 
             transformed = Cow::Owned(transformed.updated(fixed_contents, &source_map));
 
-            let tokens: Vec<LexResult> =
+            let tokens =
                 ruff_python_parser::tokenize(transformed.source_code(), source_type.as_mode());
             let locator = Locator::new(transformed.source_code());
             let stylist = Stylist::from_tokens(&tokens, &locator);
diff --git a/crates/ruff_python_parser/src/lib.rs b/crates/ruff_python_parser/src/lib.rs
index ee7a7399fdd67..75a11b6b30e7d 100644
--- a/crates/ruff_python_parser/src/lib.rs
+++ b/crates/ruff_python_parser/src/lib.rs
@@ -110,6 +110,8 @@
 //! [parsing]: https://en.wikipedia.org/wiki/Parsing
 //! [lexer]: crate::lexer
 
+use std::ops::Deref;
+
 use crate::lexer::{lex, lex_starts_at, LexResult};
 
 pub use crate::error::{FStringErrorType, ParseError, ParseErrorType};
@@ -339,8 +341,38 @@ pub fn parse_tokens(tokens: Vec<LexResult>, source: &str, mode: Mode) -> Result<
     }
 }
 
+/// Tokens represents a vector of [`LexResult`].
+#[derive(Debug, Clone)]
+pub struct Tokens(Vec<LexResult>);
+
+impl Tokens {
+    /// Returns an iterator over the [`TokenKind`] corresponding to the tokens.
+    pub fn kinds(&self) -> impl Iterator<Item = TokenKind> + '_ {
+        self.iter().flatten().map(|(tok, _)| tok.kind())
+    }
+
+    /// Consumes the [`Tokens`], returning the underlying vector of [`LexResult`].
+    pub fn into_inner(self) -> Vec<LexResult> {
+        self.0
+    }
+}
+
+impl From<Vec<LexResult>> for Tokens {
+    fn from(value: Vec<LexResult>) -> Self {
+        Self(value)
+    }
+}
+
+impl Deref for Tokens {
+    type Target = Vec<LexResult>;
+
+    fn deref(&self) -> &Self::Target {
+        &self.0
+    }
+}
+
 /// Collect tokens up to and including the first error.
-pub fn tokenize(contents: &str, mode: Mode) -> Vec<LexResult> {
+pub fn tokenize(contents: &str, mode: Mode) -> Tokens {
     let mut tokens: Vec<LexResult> = allocate_tokens_vec(contents);
     for tok in lexer::lex(contents, mode) {
         let is_err = tok.is_err();
@@ -350,7 +382,7 @@ pub fn tokenize(contents: &str, mode: Mode) -> Vec<LexResult> {
         }
     }
 
-    tokens
+    Tokens(tokens)
 }
 
 /// Tokenizes all tokens.
diff --git a/crates/ruff_python_parser/src/token.rs b/crates/ruff_python_parser/src/token.rs
index 0e1ecfaf0cbfd..e177af90de532 100644
--- a/crates/ruff_python_parser/src/token.rs
+++ b/crates/ruff_python_parser/src/token.rs
@@ -228,6 +228,11 @@ pub enum Tok {
 }
 
 impl Tok {
+    #[inline]
+    pub fn kind(&self) -> TokenKind {
+        TokenKind::from_token(self)
+    }
+
     pub fn start_marker(mode: Mode) -> Self {
         match mode {
             Mode::Module | Mode::Ipython => Tok::StartModule,
diff --git a/crates/ruff_python_trivia_integration_tests/tests/block_comments.rs b/crates/ruff_python_trivia_integration_tests/tests/block_comments.rs
index df0142b3c1798..fe6cc47ac9d7d 100644
--- a/crates/ruff_python_trivia_integration_tests/tests/block_comments.rs
+++ b/crates/ruff_python_trivia_integration_tests/tests/block_comments.rs
@@ -1,5 +1,4 @@
 use ruff_python_index::Indexer;
-use ruff_python_parser::lexer::LexResult;
 use ruff_python_parser::{tokenize, Mode};
 use ruff_source_file::Locator;
 use ruff_text_size::TextSize;
@@ -38,7 +37,7 @@ fn block_comments_indented_block() {
 fn block_comments_single_line_is_not_a_block() {
     // arrange
     let source = "\n";
-    let tokens: Vec<LexResult> = tokenize(source, Mode::Module);
+    let tokens = tokenize(source, Mode::Module);
     let locator = Locator::new(source);
     let indexer = Indexer::from_tokens(&tokens, &locator);
 
diff --git a/crates/ruff_server/src/lint.rs b/crates/ruff_server/src/lint.rs
index fdf4c54a77302..9e735302771bd 100644
--- a/crates/ruff_server/src/lint.rs
+++ b/crates/ruff_server/src/lint.rs
@@ -12,7 +12,6 @@ use ruff_linter::{
 use ruff_python_ast::PySourceType;
 use ruff_python_codegen::Stylist;
 use ruff_python_index::Indexer;
-use ruff_python_parser::lexer::LexResult;
 use ruff_python_parser::AsMode;
 use ruff_source_file::Locator;
 use ruff_text_size::Ranged;
@@ -63,7 +62,7 @@ pub(crate) fn check(
     let source_kind = SourceKind::Python(contents.to_string());
 
     // Tokenize once.
-    let tokens: Vec<LexResult> = ruff_python_parser::tokenize(contents, source_type.as_mode());
+    let tokens = ruff_python_parser::tokenize(contents, source_type.as_mode());
 
     // Map row and column locations to byte slices (lazily).
     let locator = Locator::with_index(contents, index);
diff --git a/crates/ruff_wasm/src/lib.rs b/crates/ruff_wasm/src/lib.rs
index c9dd3603e9132..56843a82e0c5a 100644
--- a/crates/ruff_wasm/src/lib.rs
+++ b/crates/ruff_wasm/src/lib.rs
@@ -17,7 +17,6 @@ use ruff_python_ast::{Mod, PySourceType};
 use ruff_python_codegen::Stylist;
 use ruff_python_formatter::{format_module_ast, pretty_comments, PyFormatContext, QuoteStyle};
 use ruff_python_index::{CommentRangesBuilder, Indexer};
-use ruff_python_parser::lexer::LexResult;
 use ruff_python_parser::{parse_tokens, tokenize_all, AsMode, Mode, Program};
 use ruff_python_trivia::CommentRanges;
 use ruff_source_file::{Locator, SourceLocation};
@@ -162,7 +161,7 @@ impl Workspace {
         let source_kind = SourceKind::Python(contents.to_string());
 
         // Tokenize once.
-        let tokens: Vec<LexResult> = ruff_python_parser::tokenize(contents, source_type.as_mode());
+        let tokens = ruff_python_parser::tokenize(contents, source_type.as_mode());
 
         // Map row and column locations to byte slices (lazily).
         let locator = Locator::new(contents);