Merge pull request #730 from parea-ai/PAI-991-new-gpt-models-py-sdk

fix(openai): support new model in costs

Author: jalexanderII

parea/constants.py

@@ -105,6 +105,11 @@ def str2bool(v):
         "completion": 60.0,
         "token_limit": {"max_completion_tokens": 4096, "max_prompt_tokens": 128000},
     },
+    "gpt-4-1106-vision-preview": {
+        "prompt": 30.0,
+        "completion": 60.0,
+        "token_limit": {"max_completion_tokens": 4096, "max_prompt_tokens": 128000},
+    },
     "gpt-4-turbo-preview": {
         "prompt": 10.0,
         "completion": 30.0,
@@ -120,6 +125,16 @@ def str2bool(v):
         "completion": 30.0,
         "token_limit": {"max_completion_tokens": 4096, "max_prompt_tokens": 128000},
     },
+    "gpt-4-turbo": {
+        "prompt": 10.0,
+        "completion": 30.0,
+        "token_limit": {"max_completion_tokens": 4096, "max_prompt_tokens": 128000},
+    },
+    "gpt-4-turbo-2024-04-09": {
+        "prompt": 10.0,
+        "completion": 30.0,
+        "token_limit": {"max_completion_tokens": 4096, "max_prompt_tokens": 128000},
+    },
 }
 AZURE_MODEL_INFO: Dict[str, Dict[str, Union[float, int, Dict[str, int]]]] = {
     "gpt-35-turbo": {
@@ -192,11 +207,26 @@ def str2bool(v):
         "completion": 30.0,
         "token_limit": {"max_completion_tokens": 4096, "max_prompt_tokens": 128000},
     },
+    "gpt-4-1106-vision-preview": {
+        "prompt": 30.0,
+        "completion": 60.0,
+        "token_limit": {"max_completion_tokens": 4096, "max_prompt_tokens": 128000},
+    },
     "gpt-35-turbo-instruct": {
         "prompt": 10.0,
         "completion": 30.0,
         "token_limit": {"max_completion_tokens": 4096, "max_prompt_tokens": 128000},
     },
+    "gpt-4-turbo": {
+        "prompt": 10.0,
+        "completion": 30.0,
+        "token_limit": {"max_completion_tokens": 4096, "max_prompt_tokens": 128000},
+    },
+    "gpt-4-turbo-2024-04-09": {
+        "prompt": 10.0,
+        "completion": 30.0,
+        "token_limit": {"max_completion_tokens": 4096, "max_prompt_tokens": 128000},
+    },
 }
 
 instant_pricing = {"prompt": 1.63, "completion": 5.51}

parea/cookbook/tracing_with_open_ai_endpoint_directly.py

@@ -17,7 +17,7 @@
 p.wrap_openai_client(client)
 
 
-def call_llm(data: List[dict], model: str = "gpt-3.5-turbo", temperature: float = 0.0) -> str:
+def call_llm(data: List[dict], model: str = "gpt-4-turbo", temperature: float = 0.0) -> str:
     return client.chat.completions.create(model=model, temperature=temperature, messages=data).choices[0].message.content
 
 
@@ -82,7 +82,7 @@ def argument_chain(query: str, additional_description: str = "") -> Tuple[str, s
 @trace
 def json_call() -> str:
     completion = client.chat.completions.create(
-        model="gpt-3.5-turbo-0125",
+        model="gpt-4-turbo-2024-04-09",
         messages=[{"role": "system", "content": "You are a helpful assistant talking in JSON."}, {"role": "user", "content": "What are you?"}],
         response_format={"type": "json_object"},
     )
@@ -95,11 +95,11 @@ def json_call() -> str:
         additional_description="Provide a concise, few sentence argument on why sparkling wine is good for you.",
     )
     print(result)
-    p.record_feedback(
-        FeedbackRequest(
-            trace_id=trace_id,
-            score=0.7,  # 0.0 (bad) to 1.0 (good)
-        )
-    )
+    # p.record_feedback(
+    #     FeedbackRequest(
+    #         trace_id=trace_id,
+    #         score=0.7,  # 0.0 (bad) to 1.0 (good)
+    #     )
+    # )
 
     print(json_call())

pyproject.toml

@@ -6,7 +6,7 @@ build-backend = "poetry.core.masonry.api"
 [tool.poetry]
 name = "parea-ai"
 packages = [{ include = "parea" }]
-version = "0.2.127"
+version = "0.2.128"
 description = "Parea python sdk"
 readme = "README.md"
 authors = ["joel-parea-ai <[email protected]>"]