feat: Add support for additional OpenAI model configurations

- Updated pyproject.toml to upgrade langchain-openai and langchain dependencies
- Modified LLMEndpoint to adjust temperature for specific models
- Added new OpenAI model configurations in LLMModelConfig:
  - o3-mini
  - o1-mini
  - o1-preview
  - o1

core/pyproject.toml

@@ -6,13 +6,13 @@ authors = [{ name = "Stan Girard", email = "[email protected]" }]
 dependencies = [
     "pydantic>=2.8.2",
     "langchain-core>=0.3,<0.4",
-    "langchain>=0.3,<0.4",
+    "langchain>=0.3.9,<0.4",
     "langgraph>=0.2.38,<0.3",
     "httpx>=0.27.0",
     "rich>=13.7.1",
     "tiktoken>=0.7.0",
     "aiofiles>=23.1.0",
-    "langchain-openai>=0.1.0",
+    "langchain-openai>=0.3.0",
     "langchain-cohere>=0.1.0",
     "langchain-community>=0.3,<0.4",
     "langchain-anthropic>=0.1.23",

core/quivr_core/llm/llm_endpoint.py

@@ -1,15 +1,15 @@
 import logging
 import os
+import time
 from typing import Union
 from urllib.parse import parse_qs, urlparse
 
 import tiktoken
 from langchain_anthropic import ChatAnthropic
-from langchain_mistralai import ChatMistralAI
 from langchain_core.language_models.chat_models import BaseChatModel
+from langchain_mistralai import ChatMistralAI
 from langchain_openai import AzureChatOpenAI, ChatOpenAI
 from pydantic import SecretStr
-import time
 
 from quivr_core.brain.info import LLMInfo
 from quivr_core.rag.entities.config import DefaultModelSuppliers, LLMEndpointConfig
@@ -242,7 +242,9 @@ def from_config(cls, config: LLMEndpointConfig = LLMEndpointConfig()):
                     else None,
                     base_url=config.llm_base_url,
                     max_tokens=config.max_output_tokens,
-                    temperature=config.temperature,
+                    temperature=config.temperature
+                    if config.model.startswith("o")
+                    else 0,
                 )
             elif config.supplier == DefaultModelSuppliers.MISTRAL:
                 _llm = ChatMistralAI(

core/quivr_core/rag/entities/config.py

@@ -1,20 +1,20 @@
+import logging
 import os
 import re
-import logging
 from enum import Enum
-from typing import Dict, Hashable, List, Optional, Union, Any, Type
+from typing import Any, Dict, Hashable, List, Optional, Type, Union
 from uuid import UUID
-from pydantic import BaseModel
-from langgraph.graph import START, END
-from langchain_core.tools import BaseTool
-from quivr_core.config import MegaparseConfig
-from rapidfuzz import process, fuzz
 
+from langchain_core.tools import BaseTool
+from langgraph.graph import END, START
+from pydantic import BaseModel
+from rapidfuzz import fuzz, process
 
 from quivr_core.base_config import QuivrBaseConfig
+from quivr_core.config import MegaparseConfig
+from quivr_core.llm_tools.llm_tools import TOOLS_CATEGORIES, TOOLS_LISTS, LLMToolFactory
 from quivr_core.processor.splitter import SplitterConfig
 from quivr_core.rag.prompts import CustomPromptsModel
-from quivr_core.llm_tools.llm_tools import LLMToolFactory, TOOLS_CATEGORIES, TOOLS_LISTS
 
 logger = logging.getLogger("quivr_core")
 
@@ -93,6 +93,26 @@ class LLMModelConfig:
                 max_output_tokens=16384,
                 tokenizer_hub="Quivr/gpt-4o",
             ),
+            "o3-mini": LLMConfig(
+                max_context_tokens=200000,
+                max_output_tokens=100000,
+                tokenizer_hub="Quivr/gpt-4o",
+            ),
+            "o1-mini": LLMConfig(
+                max_context_tokens=128000,
+                max_output_tokens=65536,
+                tokenizer_hub="Quivr/gpt-4o",
+            ),
+            "o1-preview": LLMConfig(
+                max_context_tokens=128000,
+                max_output_tokens=32768,
+                tokenizer_hub="Quivr/gpt-4o",
+            ),
+            "o1": LLMConfig(
+                max_context_tokens=200000,
+                max_output_tokens=100000,
+                tokenizer_hub="Quivr/gpt-4o",
+            ),
             "gpt-4-turbo": LLMConfig(
                 max_context_tokens=128000,
                 max_output_tokens=4096,