Add support for o3 mini model by openai

src/khoj/processor/conversation/openai/utils.py

@@ -71,6 +71,8 @@ def completion_with_backoff(
     elif model_name.startswith("o1"):
         temperature = 1
         model_kwargs.pop("response_format", None)
+    elif model_name.startswith("o3-"):
+        temperature = 1
 
     if os.getenv("KHOJ_LLM_SEED"):
         model_kwargs["seed"] = int(os.getenv("KHOJ_LLM_SEED"))
@@ -181,6 +183,8 @@ def llm_thread(
         elif model_name.startswith("o1-"):
             temperature = 1
             model_kwargs.pop("response_format", None)
+        elif model_name.startswith("o3-"):
+            temperature = 1
         elif model_name.startswith("deepseek-reasoner"):
             # Two successive messages cannot be from the same role. Should merge any back-to-back messages from the same role.
             # The first message should always be a user message (except system message).

src/khoj/processor/conversation/utils.py

@@ -55,6 +55,7 @@ model_to_prompt_size = {
     "gpt-4o-mini": 60000,
     "o1": 20000,
     "o1-mini": 60000,
+    "o3-mini": 60000,
     # Google Models
     "gemini-1.5-flash": 60000,
     "gemini-1.5-pro": 60000,

src/khoj/utils/constants.py

@@ -40,6 +40,7 @@ model_to_cost: Dict[str, Dict[str, float]] = {
     "gpt-4o-mini": {"input": 0.15, "output": 0.60},
     "o1": {"input": 15.0, "output": 60.00},
     "o1-mini": {"input": 3.0, "output": 12.0},
+    "o3-mini": {"input": 1.10, "output": 4.40},
     # Gemini Pricing: https://ai.google.dev/pricing
     "gemini-1.5-flash": {"input": 0.075, "output": 0.30},
     "gemini-1.5-flash-002": {"input": 0.075, "output": 0.30},