diff --git a/pyproject.toml b/pyproject.toml
index e9faaa8..7ef2295 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -23,7 +23,7 @@ requires = ["hatchling"]
 build-backend = "hatchling.build"
 
 [tool.uv]
-dev-dependencies = ["pytest>=8.3.2", "codecov>=2.1.13"]
+dev-dependencies = ["pytest>=8.3.2", "pytest-vcr>=1.0.2", "codecov>=2.1.13"]
 
 [project.entry-points."exchange.provider"]
 openai = "exchange.providers.openai:OpenAiProvider"
diff --git a/tests/providers/openai/__init__.py b/tests/providers/openai/__init__.py
new file mode 100644
index 0000000..8c5df76
--- /dev/null
+++ b/tests/providers/openai/__init__.py
@@ -0,0 +1 @@
+"""Tests that use the OpenAI API."""
diff --git a/tests/providers/openai/cassettes/test_ollama_completion.yaml b/tests/providers/openai/cassettes/test_ollama_completion.yaml
new file mode 100644
index 0000000..88bc206
--- /dev/null
+++ b/tests/providers/openai/cassettes/test_ollama_completion.yaml
@@ -0,0 +1,68 @@
+interactions:
+- request:
+    body: ''
+    headers:
+      accept:
+      - '*/*'
+      accept-encoding:
+      - gzip, deflate
+      connection:
+      - keep-alive
+      host:
+      - localhost:11434
+      user-agent:
+      - python-httpx/0.27.2
+    method: GET
+    uri: http://localhost:11434/
+  response:
+    body:
+      string: Ollama is running
+    headers:
+      Content-Length:
+      - '17'
+      Content-Type:
+      - text/plain; charset=utf-8
+      Date:
+      - Sun, 22 Sep 2024 23:40:13 GMT
+      Set-Cookie: test_set_cookie
+      openai-organization: test_openai_org_key
+    status:
+      code: 200
+      message: OK
+- request:
+    body: '{"messages": [{"role": "system", "content": "You are a helpful assistant."},
+      {"role": "user", "content": "Hello"}], "model": "mistral-nemo"}'
+    headers:
+      accept:
+      - '*/*'
+      accept-encoding:
+      - gzip, deflate
+      connection:
+      - keep-alive
+      content-length:
+      - '140'
+      content-type:
+      - application/json
+      host:
+      - localhost:11434
+      user-agent:
+      - python-httpx/0.27.2
+    method: POST
+    uri: http://localhost:11434/v1/chat/completions
+  response:
+    body:
+      string: "{\"id\":\"chatcmpl-429\",\"object\":\"chat.completion\",\"created\":1727048416,\"model\":\"mistral-nemo\",\"system_fingerprint\":\"fp_ollama\",\"choices\":[{\"index\":0,\"message\":{\"role\":\"assistant\",\"content\":\"Hello!
+        I'm here to help. How can I assist you today? Let's chat. \U0001F60A\"},\"finish_reason\":\"stop\"}],\"usage\":{\"prompt_tokens\":10,\"completion_tokens\":23,\"total_tokens\":33}}\n"
+    headers:
+      Content-Length:
+      - '356'
+      Content-Type:
+      - application/json
+      Date:
+      - Sun, 22 Sep 2024 23:40:16 GMT
+      Set-Cookie: test_set_cookie
+      openai-organization: test_openai_org_key
+    status:
+      code: 200
+      message: OK
+version: 1
diff --git a/tests/providers/openai/cassettes/test_openai_completion.yaml b/tests/providers/openai/cassettes/test_openai_completion.yaml
new file mode 100644
index 0000000..1a92eb3
--- /dev/null
+++ b/tests/providers/openai/cassettes/test_openai_completion.yaml
@@ -0,0 +1,80 @@
+interactions:
+- request:
+    body: '{"messages": [{"role": "system", "content": "You are a helpful assistant."},
+      {"role": "user", "content": "Hello"}], "model": "gpt-4o-mini"}'
+    headers:
+      accept:
+      - '*/*'
+      accept-encoding:
+      - gzip, deflate
+      authorization:
+      - Bearer test_openai_api_key
+      connection:
+      - keep-alive
+      content-length:
+      - '139'
+      content-type:
+      - application/json
+      host:
+      - api.openai.com
+      user-agent:
+      - python-httpx/0.27.2
+    method: POST
+    uri: https://api.openai.com/v1/chat/completions
+  response:
+    body:
+      string: "{\n  \"id\": \"chatcmpl-AAQTYi3DXJnltAfd5sUH1Wnzh69t3\",\n  \"object\":
+        \"chat.completion\",\n  \"created\": 1727048416,\n  \"model\": \"gpt-4o-mini-2024-07-18\",\n
+        \ \"choices\": [\n    {\n      \"index\": 0,\n      \"message\": {\n        \"role\":
+        \"assistant\",\n        \"content\": \"Hello! How can I assist you today?\",\n
+        \       \"refusal\": null\n      },\n      \"logprobs\": null,\n      \"finish_reason\":
+        \"stop\"\n    }\n  ],\n  \"usage\": {\n    \"prompt_tokens\": 18,\n    \"completion_tokens\":
+        9,\n    \"total_tokens\": 27,\n    \"completion_tokens_details\": {\n      \"reasoning_tokens\":
+        0\n    }\n  },\n  \"system_fingerprint\": \"fp_1bb46167f9\"\n}\n"
+    headers:
+      CF-Cache-Status:
+      - DYNAMIC
+      CF-RAY:
+      - 8c762399feb55739-SYD
+      Connection:
+      - keep-alive
+      Content-Type:
+      - application/json
+      Date:
+      - Sun, 22 Sep 2024 23:40:17 GMT
+      Server:
+      - cloudflare
+      Set-Cookie: test_set_cookie
+      Transfer-Encoding:
+      - chunked
+      X-Content-Type-Options:
+      - nosniff
+      access-control-expose-headers:
+      - X-Request-ID
+      content-length:
+      - '593'
+      openai-organization: test_openai_org_key
+      openai-processing-ms:
+      - '560'
+      openai-version:
+      - '2020-10-01'
+      strict-transport-security:
+      - max-age=15552000; includeSubDomains; preload
+      x-ratelimit-limit-requests:
+      - '10000'
+      x-ratelimit-limit-tokens:
+      - '200000'
+      x-ratelimit-remaining-requests:
+      - '9999'
+      x-ratelimit-remaining-tokens:
+      - '199973'
+      x-ratelimit-reset-requests:
+      - 8.64s
+      x-ratelimit-reset-tokens:
+      - 8ms
+      x-request-id:
+      - req_22e26c840219cde3152eaba1ce89483b
+    status:
+      code: 200
+      message: OK
+version: 1
diff --git a/tests/providers/openai/conftest.py b/tests/providers/openai/conftest.py
new file mode 100644
index 0000000..e282c87
--- /dev/null
+++ b/tests/providers/openai/conftest.py
@@ -0,0 +1,52 @@
+import os
+import pytest
+
+OPENAI_MODEL = "gpt-4o-mini"
+OPENAI_API_KEY = "test_openai_api_key"
+OPENAI_ORG_ID = "test_openai_org_key"
+OPENAI_PROJECT_ID = "test_openai_project_id"
+
+
+@pytest.fixture
+def default_openai_api_key(monkeypatch):
+    """
+    This fixture avoids the error OpenAiProvider.from_env() raises when the
+    OPENAI_API_KEY is not set in the environment.
+
+    When running VCR tests for the first time or after deleting a cassette
+    recording, a real OPENAI_API_KEY must be passed as an environment variable,
+    so real responses can be fetched. Subsequent runs use the recorded data, so
+    don't need a real key.
+    """
+    if "OPENAI_API_KEY" not in os.environ:
+        monkeypatch.setenv("OPENAI_API_KEY", OPENAI_API_KEY)
+
+
+@pytest.fixture(scope="module")
+def vcr_config():
+    """
+    This scrubs sensitive data and gunzips bodies when in recording mode.
+
+    Without this, you would leak cookies and auth tokens in the cassettes.
+    Also, depending on the request, some responses would be binary encoded
+    while others plain json. This ensures all bodies are human-readable.
+    """
+    return {
+        "decode_compressed_response": True,
+        "filter_headers": [
+            ("authorization", "Bearer " + OPENAI_API_KEY),
+            ("openai-organization", OPENAI_ORG_ID),
+            ("openai-project", OPENAI_PROJECT_ID),
+            ("cookie", None),
+        ],
+        "before_record_response": scrub_response_headers,
+    }
+
+
+def scrub_response_headers(response):
+    """
+    This scrubs sensitive response headers. Note they are case-sensitive!
+    """
+    response["headers"]["openai-organization"] = OPENAI_ORG_ID
+    response["headers"]["Set-Cookie"] = "test_set_cookie"
+    return response
diff --git a/tests/providers/openai/test_ollama.py b/tests/providers/openai/test_ollama.py
new file mode 100644
index 0000000..a432d0c
--- /dev/null
+++ b/tests/providers/openai/test_ollama.py
@@ -0,0 +1,32 @@
+from typing import Tuple
+
+import pytest
+
+from exchange import Text
+from exchange.message import Message
+from exchange.providers.base import Usage
+from exchange.providers.ollama import OllamaProvider, OLLAMA_MODEL
+
+
+@pytest.mark.vcr()
+def test_ollama_completion(default_openai_api_key):
+    reply_message, reply_usage = ollama_complete()
+
+    assert reply_message.content == [Text(text="Hello! I'm here to help. How can I assist you today? Let's chat. 😊")]
+    assert reply_usage.total_tokens == 33
+
+
+@pytest.mark.integration
+def test_ollama_completion_integration():
+    reply = ollama_complete()
+
+    assert reply[0].content is not None
+    print("Completion content from OpenAI:", reply[0].content)
+
+
+def ollama_complete() -> Tuple[Message, Usage]:
+    provider = OllamaProvider.from_env()
+    model = OLLAMA_MODEL
+    system = "You are a helpful assistant."
+    messages = [Message.user("Hello")]
+    return provider.complete(model=model, system=system, messages=messages, tools=None)
diff --git a/tests/providers/openai/test_openai.py b/tests/providers/openai/test_openai.py
new file mode 100644
index 0000000..354d1c5
--- /dev/null
+++ b/tests/providers/openai/test_openai.py
@@ -0,0 +1,39 @@
+from typing import Tuple
+
+import os
+import pytest
+
+from exchange import Text
+from exchange.message import Message
+from exchange.providers.base import Usage
+from exchange.providers.openai import OpenAiProvider
+from .conftest import OPENAI_MODEL, OPENAI_API_KEY
+
+
+@pytest.mark.vcr()
+def test_openai_completion(monkeypatch):
+    # When running VCR tests the first time, it needs OPENAI_API_KEY to call
+    # the real service. Afterward, it is not needed as VCR mocks the service.
+    if "OPENAI_API_KEY" not in os.environ:
+        monkeypatch.setenv("OPENAI_API_KEY", OPENAI_API_KEY)
+
+    reply_message, reply_usage = openai_complete()
+
+    assert reply_message.content == [Text(text="Hello! How can I assist you today?")]
+    assert reply_usage.total_tokens == 27
+
+
+@pytest.mark.integration
+def test_openai_completion_integration():
+    reply = openai_complete()
+
+    assert reply[0].content is not None
+    print("Completion content from OpenAI:", reply[0].content)
+
+
+def openai_complete() -> Tuple[Message, Usage]:
+    provider = OpenAiProvider.from_env()
+    model = OPENAI_MODEL
+    system = "You are a helpful assistant."
+    messages = [Message.user("Hello")]
+    return provider.complete(model=model, system=system, messages=messages, tools=None)
diff --git a/tests/providers/test_ollama.py b/tests/providers/test_ollama.py
deleted file mode 100644
index 7812fe6..0000000
--- a/tests/providers/test_ollama.py
+++ /dev/null
@@ -1,16 +0,0 @@
-import pytest
-from exchange import Message
-from exchange.providers.ollama import OllamaProvider, OLLAMA_MODEL
-
-
-@pytest.mark.integration
-def test_ollama_integration():
-    provider = OllamaProvider.from_env()
-    model = OLLAMA_MODEL
-    system = "You are a helpful assistant."
-    messages = [Message.user("Hello")]
-
-    reply = provider.complete(model=model, system=system, messages=messages, tools=None)
-
-    assert reply[0].content is not None
-    print("Completion content from Ollama:", reply[0].content)
diff --git a/tests/providers/test_openai.py b/tests/providers/test_openai.py
deleted file mode 100644
index 0e0e000..0000000
--- a/tests/providers/test_openai.py
+++ /dev/null
@@ -1,58 +0,0 @@
-import os
-from unittest.mock import patch
-
-import pytest
-from exchange import Message, Text
-from exchange.providers.openai import OpenAiProvider
-
-
-@pytest.fixture
-@patch.dict(os.environ, {"OPENAI_API_KEY": "test_api_key"})
-def openai_provider():
-    return OpenAiProvider.from_env()
-
-
-@patch("httpx.Client.post")
-@patch("time.sleep", return_value=None)
-@patch("logging.warning")
-@patch("logging.error")
-def test_openai_completion(mock_error, mock_warning, mock_sleep, mock_post, openai_provider):
-    mock_response = {
-        "choices": [{"message": {"role": "assistant", "content": "Hello!"}}],
-        "usage": {"prompt_tokens": 10, "completion_tokens": 25, "total_tokens": 35},
-    }
-
-    mock_post.return_value.json.return_value = mock_response
-
-    model = "gpt-4"
-    system = "You are a helpful assistant."
-    messages = [Message.user("Hello")]
-    tools = ()
-
-    reply_message, reply_usage = openai_provider.complete(model=model, system=system, messages=messages, tools=tools)
-
-    assert reply_message.content == [Text(text="Hello!")]
-    assert reply_usage.total_tokens == 35
-    mock_post.assert_called_once_with(
-        "v1/chat/completions",
-        json={
-            "messages": [
-                {"role": "system", "content": system},
-                {"role": "user", "content": "Hello"},
-            ],
-            "model": model,
-        },
-    )
-
-
-@pytest.mark.integration
-def test_openai_integration():
-    provider = OpenAiProvider.from_env()
-    model = "gpt-4"  # specify a valid model
-    system = "You are a helpful assistant."
-    messages = [Message.user("Hello")]
-
-    reply = provider.complete(model=model, system=system, messages=messages, tools=None)
-
-    assert reply[0].content is not None
-    print("Completion content from OpenAI:", reply[0].content)