refactor: rename max_tokens to max_completion_tokens for SDK v5 compatibility

- Updated parameter name in index.js to align with SDK v5 specifications.
- Adjusted test case to verify the new parameter name and ensure deprecated parameter is not present.

Author: ryota-murakami

index.js

@@ -158,7 +158,7 @@ Commit message: `,
     model,
     messages,
     temperature: 0.7,
-    max_tokens: 200,
+    max_completion_tokens: 200,
   }
 
   let response

tests/gpt5-models.test.js

@@ -97,16 +97,16 @@ describe('Model Configuration', () => {
       )
     })
 
-    it('should use max_tokens parameter (SDK v5 compatible)', async () => {
+    it('should use max_completion_tokens parameter (forward-compatible)', async () => {
       const indexContent = fs.readFileSync(
         path.join(__dirname, '..', 'index.js'),
         'utf8',
       )
 
-      // Check that max_tokens is used (SDK v5 parameter)
-      expect(indexContent).toContain('max_tokens: 200')
+      // Check that max_completion_tokens is used (works with all models including o1)
+      expect(indexContent).toContain('max_completion_tokens: 200')
       // Should NOT contain deprecated parameter
-      expect(indexContent).not.toContain('max_completion_tokens')
+      expect(indexContent).not.toContain('max_tokens:')
     })
   })