diff --git a/packages/instrumentation-openai/README.md b/packages/instrumentation-openai/README.md
index 725897dd..288b3040 100644
--- a/packages/instrumentation-openai/README.md
+++ b/packages/instrumentation-openai/README.md
@@ -126,8 +126,8 @@ ollama serve
 
 # When using Ollama, we default to qwen2.5:0.5b, which is a small model. You
 # can choose a larger one, or a different tool capable model like mistral-nemo.
-export MODEL_CHAT=qwen2.5
-ollama pull $MODEL_CHAT
+export CHAT_MODEL=qwen2.5
+ollama pull $CHAT_MODEL
 
 OPENAI_BASE_URL=http://localhost:11434/v1 \
     node use-chat.js
diff --git a/packages/instrumentation-openai/examples/openai.js b/packages/instrumentation-openai/examples/openai.js
index c63ea2da..d7ea72c3 100644
--- a/packages/instrumentation-openai/examples/openai.js
+++ b/packages/instrumentation-openai/examples/openai.js
@@ -35,9 +35,9 @@ function newOpenAIAndModels() {
   // Default to models available in both the OpenAI platform and Azure OpenAI
   // Service. For Azure, however, this "model" must match the Azure "deployment
   // name".
-  let chatModel = process.env.MODEL_CHAT ?? 'gpt-4o-mini';
+  let chatModel = process.env.CHAT_MODEL ?? 'gpt-4o-mini';
   let embeddingsModel =
-    process.env.MODEL_EMBEDDINGS ?? 'text-embedding-3-small';
+    process.env.EMBEDDINGS_MODEL ?? 'text-embedding-3-small';
 
   if (process.env.AZURE_OPENAI_API_KEY) {
     clientCtor = AzureOpenAI;
@@ -47,8 +47,8 @@ function newOpenAIAndModels() {
   ) {
     process.env.OPENAI_API_KEY = 'unused';
     // Note: Others like LocalAI do not use Ollama's naming scheme.
-    chatModel = process.env.MODEL_CHAT ?? 'qwen2.5:0.5b';
-    embeddingsModel = process.env.MODEL_EMBEDDINGS ?? 'all-minilm:33m';
+    chatModel = process.env.CHAT_MODEL ?? 'qwen2.5:0.5b';
+    embeddingsModel = process.env.EMBEDDINGS_MODEL ?? 'all-minilm:33m';
   }
 
   return { client: new clientCtor(), chatModel, embeddingsModel };
diff --git a/packages/instrumentation-openai/test/fixtures.test.js b/packages/instrumentation-openai/test/fixtures.test.js
index dffce81f..164d2868 100644
--- a/packages/instrumentation-openai/test/fixtures.test.js
+++ b/packages/instrumentation-openai/test/fixtures.test.js
@@ -107,8 +107,8 @@ if (process.env.TEST_FIXTURES_ENV_FILE) {
   require('dotenv').config({ path: process.env.TEST_FIXTURES_ENV_FILE });
 }
 
-const UNIT_TEST_MODEL_TOOLS = 'gpt-4o-mini';
-const UNIT_TEST_MODEL_EMBEDDINGS = 'text-embedding-3-small';
+const UNIT_TEST_CHAT_MODEL = 'gpt-4o-mini';
+const UNIT_TEST_EMBEDDINGS_MODEL = 'text-embedding-3-small';
 
 // Configure the test fixtures based on the test mode.
 const testMode = process.env.TEST_FIXTURES_MODE || 'unit';
@@ -130,8 +130,8 @@ switch (testMode) {
     // OPENAI_API_KEY needs to be set to something to avoid OpenAI
     // constructor error. However, because of mocking, it isn't used.
     process.env.OPENAI_API_KEY = 'notused';
-    process.env.TEST_CHAT_MODEL = UNIT_TEST_MODEL_TOOLS;
-    process.env.TEST_EMBEDDINGS_MODEL = UNIT_TEST_MODEL_EMBEDDINGS;
+    process.env.TEST_CHAT_MODEL = UNIT_TEST_CHAT_MODEL;
+    process.env.TEST_EMBEDDINGS_MODEL = UNIT_TEST_EMBEDDINGS_MODEL;
     targetService = 'openai';
     break;
 
@@ -149,8 +149,8 @@ switch (testMode) {
     }
     usingNock = true;
     process.env.TEST_NOCK_BACK_MODE = 'update';
-    process.env.TEST_CHAT_MODEL = UNIT_TEST_MODEL_TOOLS;
-    process.env.TEST_EMBEDDINGS_MODEL = UNIT_TEST_MODEL_EMBEDDINGS;
+    process.env.TEST_CHAT_MODEL = UNIT_TEST_CHAT_MODEL;
+    process.env.TEST_EMBEDDINGS_MODEL = UNIT_TEST_EMBEDDINGS_MODEL;
     targetService = 'openai';
     break;