Merge branch 'master' of github.com:khoj-ai/khoj into features/update-home-page

src/interface/web/app/common/layoutHelper.tsx

@@ -9,7 +9,7 @@ export function ContentSecurityPolicy() {
                style-src 'self' https://assets.khoj.dev 'unsafe-inline' https://fonts.googleapis.com https://accounts.google.com;
                img-src 'self' data: blob: https://*.khoj.dev https://accounts.google.com https://*.googleusercontent.com https://*.google.com/ https://*.gstatic.com;
                font-src 'self' https://assets.khoj.dev https://fonts.gstatic.com;
-               frame-src 'self' https://accounts.google.com;
+               frame-src 'self' https://accounts.google.com https://app.chatwoot.com;
                child-src 'self' https://app.chatwoot.com;
                object-src 'none';"
         ></meta>

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

@@ -58,13 +58,17 @@ def completion_with_backoff(
         openai_clients[client_key] = client
 
     formatted_messages = [{"role": message.role, "content": message.content} for message in messages]
-    stream = True
 
     # Update request parameters for compatability with o1 model series
     # Refer: https://platform.openai.com/docs/guides/reasoning/beta-limitations
-    if model_name.startswith("o1"):
+    stream = True
+    model_kwargs["stream_options"] = {"include_usage": True}
+    if model_name == "o1":
+        temperature = 1
+        stream = False
+        model_kwargs.pop("stream_options", None)
+    elif model_name.startswith("o1"):
         temperature = 1
-        model_kwargs.pop("stop", None)
         model_kwargs.pop("response_format", None)
 
     if os.getenv("KHOJ_LLM_SEED"):
@@ -74,7 +78,6 @@ def completion_with_backoff(
         messages=formatted_messages,  # type: ignore
         model=model_name,  # type: ignore
         stream=stream,
-        stream_options={"include_usage": True} if stream else {},
         temperature=temperature,
         timeout=20,
         **model_kwargs,
@@ -165,13 +168,17 @@ def llm_thread(
             client = openai_clients[client_key]
 
         formatted_messages = [{"role": message.role, "content": message.content} for message in messages]
-        stream = True
 
         # Update request parameters for compatability with o1 model series
         # Refer: https://platform.openai.com/docs/guides/reasoning/beta-limitations
-        if model_name.startswith("o1"):
+        stream = True
+        model_kwargs["stream_options"] = {"include_usage": True}
+        if model_name == "o1":
+            temperature = 1
+            stream = False
+            model_kwargs.pop("stream_options", None)
+        elif model_name.startswith("o1-"):
             temperature = 1
-            model_kwargs.pop("stop", None)
             model_kwargs.pop("response_format", None)
 
         if os.getenv("KHOJ_LLM_SEED"):
@@ -181,7 +188,6 @@ def llm_thread(
             messages=formatted_messages,
             model=model_name,  # type: ignore
             stream=stream,
-            stream_options={"include_usage": True} if stream else {},
             temperature=temperature,
             timeout=20,
             **model_kwargs,

src/khoj/processor/conversation/utils.py

@@ -34,7 +34,6 @@ from khoj.search_filter.word_filter import WordFilter
 from khoj.utils import state
 from khoj.utils.helpers import (
     ConversationCommand,
-    in_debug_mode,
     is_none_or_empty,
     is_promptrace_enabled,
     merge_dicts,
@@ -47,28 +46,27 @@ logger = logging.getLogger(__name__)
 try:
     from git import Repo
 except ImportError:
-    if in_debug_mode():
-        logger.warning("GitPython not installed. `pip install gitpython` to enable prompt tracer.")
+    if is_promptrace_enabled():
+        logger.warning("GitPython not installed. `pip install gitpython` to use prompt tracer.")
 
 model_to_prompt_size = {
     # OpenAI Models
-    "gpt-4o": 20000,
-    "gpt-4o-mini": 20000,
-    "o1-preview": 20000,
-    "o1-mini": 20000,
+    "gpt-4o": 60000,
+    "gpt-4o-mini": 60000,
+    "o1": 20000,
+    "o1-mini": 60000,
     # Google Models
-    "gemini-1.5-flash": 20000,
-    "gemini-1.5-pro": 20000,
+    "gemini-1.5-flash": 60000,
+    "gemini-1.5-pro": 60000,
     # Anthropic Models
-    "claude-3-5-sonnet-20241022": 20000,
-    "claude-3-5-haiku-20241022": 20000,
+    "claude-3-5-sonnet-20241022": 60000,
+    "claude-3-5-haiku-20241022": 60000,
     # Offline Models
-    "bartowski/Meta-Llama-3.1-8B-Instruct-GGUF": 20000,
+    "Qwen/Qwen2.5-14B-Instruct-GGUF": 20000,
     "bartowski/Meta-Llama-3.1-8B-Instruct-GGUF": 20000,
     "bartowski/Llama-3.2-3B-Instruct-GGUF": 20000,
     "bartowski/gemma-2-9b-it-GGUF": 6000,
     "bartowski/gemma-2-2b-it-GGUF": 6000,
-    "Qwen/Qwen2.5-14B-Instruct-GGUF": 20000,
 }
 model_to_tokenizer: Dict[str, str] = {}
 

src/khoj/utils/constants.py

@@ -38,7 +38,7 @@ model_to_cost: Dict[str, Dict[str, float]] = {
     # OpenAI Pricing: https://openai.com/api/pricing/
     "gpt-4o": {"input": 2.50, "output": 10.00},
     "gpt-4o-mini": {"input": 0.15, "output": 0.60},
-    "o1-preview": {"input": 15.0, "output": 60.00},
+    "o1": {"input": 15.0, "output": 60.00},
     "o1-mini": {"input": 3.0, "output": 12.0},
     # Gemini Pricing: https://ai.google.dev/pricing
     "gemini-1.5-flash": {"input": 0.075, "output": 0.30},