fix: remove max token limit from llm calls

VERSION

@@ -1 +1 @@
-0.1.0+44
+0.1.0+45

apinex_client.py

@@ -443,7 +443,6 @@ async def chat_text(
         "model": model,
         "messages": [{"role": "user", "content": user_content}],
         "temperature": 0.3,
-        "max_tokens": 1800,
     }
     payload = {
         "model": model,
@@ -452,7 +451,6 @@ async def chat_text(
             {"role": "user", "content": user_content},
         ],
         "temperature": 0.3,
-        "max_tokens": 1800,
     }
     for attempt in range(3):
         attempt_payload = dict(payload)

tests/test_writer_features.py

@@ -124,6 +124,7 @@ class ApinexTests(unittest.IsolatedAsyncioTestCase):
         _, kwargs = fake_session.request
         self.assertEqual(kwargs["json"]["model"], "gemini-3.8-flash")
         self.assertEqual(kwargs["json"]["messages"][0]["role"], "system")
+        self.assertNotIn("max_tokens", kwargs["json"])
         logs = user_store.list_api_call_logs(limit=1)
         self.assertEqual(len(logs), 1)
         log = user_store.get_api_call_log(logs[0]["id"] if isinstance(logs[0]["id"], int) else -1)
@@ -273,10 +274,10 @@ class ApinexTests(unittest.IsolatedAsyncioTestCase):
         self.assertEqual(detail["status_code"], 200)
         self.assertIsInstance(detail["duration_ms"], int)
 
-    async def test_api_log_redacts_secrets_but_keeps_token_limits(self):
+    async def test_api_log_redacts_secrets_without_token_limit(self):
         await apinex_client._record_api_call(
             stage="test", model="model", prompt_name="test", system_prompt="",
-            request={"max_tokens": 1800, "api_key": "private-key"},
+            request={"api_key": "private-key"},
             response={"authorization": "Bearer private-token", "detail": "Bearer private-token"},
         )
         log = user_store.list_api_call_logs(limit=1)[0]
@@ -286,7 +287,7 @@ class ApinexTests(unittest.IsolatedAsyncioTestCase):
         assert detail is not None
         request = json.loads(str(detail["request_json"]))
         response = json.loads(str(detail["response_json"]))
-        self.assertEqual(request["max_tokens"], 1800)
+        self.assertNotIn("max_tokens", request)
         self.assertEqual(request["api_key"], "[REDACTED]")
         self.assertEqual(response["authorization"], "[REDACTED]")
         self.assertEqual(response["detail"], "Bearer [REDACTED]")