diff --git a/src/nnbench/__init__.py b/src/nnbench/__init__.py
index 0315052..fc35ed9 100644
--- a/src/nnbench/__init__.py
+++ b/src/nnbench/__init__.py
@@ -2,7 +2,7 @@
 
 from .core import benchmark, parametrize, product
 from .reporter import BenchmarkReporter, ConsoleReporter, FileReporter
-from .runner import BenchmarkRunner
+from .runner import collect, run
 from .types import Benchmark, BenchmarkRecord, Memo, Parameters
 
 __version__ = "0.4.0"
diff --git a/src/nnbench/cli.py b/src/nnbench/cli.py
index bee2d4a..a90cea8 100644
--- a/src/nnbench/cli.py
+++ b/src/nnbench/cli.py
@@ -6,7 +6,7 @@
 import sys
 from typing import Any
 
-from nnbench import BenchmarkRunner, ConsoleReporter, __version__
+from nnbench import ConsoleReporter, __version__, collect, run
 from nnbench.config import NNBenchConfig, parse_nnbench_config
 from nnbench.reporter import FileReporter
 
@@ -205,9 +205,9 @@ def main() -> int:
                 else:
                     context[k] = v
 
-            record = BenchmarkRunner().run(
-                args.benchmarks,
-                tags=tuple(args.tags),
+            benchmarks = collect(args.benchmarks, tags=tuple(args.tags))
+            record = run(
+                benchmarks,
                 context=[lambda: context],
             )
 
diff --git a/src/nnbench/context.py b/src/nnbench/context.py
index 27a3c33..be8cfb2 100644
--- a/src/nnbench/context.py
+++ b/src/nnbench/context.py
@@ -29,7 +29,7 @@ class PythonInfo:
 
     Parameters
     ----------
-    packages: str
+    packages: Sequence[str]
         Names of the requested packages under which they exist in the current environment.
         For packages installed through ``pip``, this equals the PyPI package name.
     """
@@ -37,7 +37,7 @@ class PythonInfo:
     key = "python"
 
     def __init__(self, packages: Sequence[str] = ()):
-        self.packages = packages
+        self.packages = tuple(packages)
 
     def __call__(self) -> dict[str, Any]:
         from importlib.metadata import PackageNotFoundError, version
diff --git a/tests/test_runner.py b/tests/test_runner.py
index c1aa2ca..3c138cc 100644
--- a/tests/test_runner.py
+++ b/tests/test_runner.py
@@ -7,45 +7,29 @@
 
 
 def test_runner_collection(testfolder: str) -> None:
-    r = nnbench.BenchmarkRunner()
+    benchmarks = nnbench.collect(os.path.join(testfolder, "standard.py"), tags=("runner-collect",))
+    assert len(benchmarks) == 1
 
-    r.collect(os.path.join(testfolder, "standard.py"), tags=("runner-collect",))
-    assert len(r.benchmarks) == 1
-    r.clear()
+    benchmarks = nnbench.collect(testfolder, tags=("non-existing-tag",))
+    assert len(benchmarks) == 0
 
-    r.collect(testfolder, tags=("non-existing-tag",))
-    assert len(r.benchmarks) == 0
-    r.clear()
-
-    r.collect(testfolder, tags=("runner-collect",))
-    assert len(r.benchmarks) == 1
+    benchmarks = nnbench.collect(testfolder, tags=("runner-collect",))
+    assert len(benchmarks) == 1
 
 
 def test_tag_selection(testfolder: str) -> None:
     PATH = os.path.join(testfolder, "tags.py")
 
-    r = nnbench.BenchmarkRunner()
-
-    r.collect(PATH)
-    assert len(r.benchmarks) == 3
-    r.clear()
-
-    r.collect(PATH, tags=("tag1",))
-    assert len(r.benchmarks) == 2
-    r.clear()
-
-    r.collect(PATH, tags=("tag2",))
-    assert len(r.benchmarks) == 1
-    r.clear()
+    assert len(nnbench.collect(PATH)) == 3
+    assert len(nnbench.collect(PATH, tags=("tag1",))) == 2
+    assert len(nnbench.collect(PATH, tags=("tag2",))) == 1
 
 
 def test_context_assembly(testfolder: str) -> None:
-    r = nnbench.BenchmarkRunner()
-
     context_providers = [system, cpuarch, python_version]
-    result = r.run(
-        testfolder,
-        tags=("standard",),
+    benchmarks = nnbench.collect(testfolder, tags=("standard",))
+    result = nnbench.run(
+        benchmarks,
         params={"x": 1, "y": 1},
         context=context_providers,
     )
@@ -57,17 +41,15 @@ def test_context_assembly(testfolder: str) -> None:
 
 
 def test_error_on_duplicate_context_keys_in_runner(testfolder: str) -> None:
-    r = nnbench.BenchmarkRunner()
-
     def duplicate_context_provider() -> dict[str, str]:
         return {"system": "DuplicateSystem"}
 
     context_providers = [system, duplicate_context_provider]
 
+    benchmarks = nnbench.collect(testfolder, tags=("standard",))
     with pytest.raises(ValueError, match="got multiple values for context key 'system'"):
-        r.run(
-            testfolder,
-            tags=("standard",),
+        nnbench.run(
+            benchmarks,
             params={"x": 1, "y": 1},
             context=context_providers,
         )
@@ -78,11 +60,9 @@ def test_filter_benchmarks_on_params(testfolder: str) -> None:
     def prod(a: int, b: int = 1) -> int:
         return a * b
 
-    r = nnbench.BenchmarkRunner()
-    r.benchmarks.append(prod)
-    # TODO (nicholasjng): This is hacky
-    rec1 = r.run("", params={"a": 1, "b": 2})
+    benchmarks = [prod]
+    rec1 = nnbench.run(benchmarks, params={"a": 1, "b": 2})
     assert rec1.benchmarks[0]["parameters"] == {"a": 1, "b": 2}
     # Assert that the defaults are also present if not overridden.
-    rec2 = r.run("", params={"a": 1})
+    rec2 = nnbench.run(benchmarks, params={"a": 1})
     assert rec2.benchmarks[0]["parameters"] == {"a": 1, "b": 1}