khoj 2.0.0b12__py3-none-any.whl → 2.0.0b13__py3-none-any.whl

khoj/routers/helpers.py

@@ -1,6 +1,5 @@
 import asyncio
 import base64
-import concurrent.futures
 import fnmatch
 import hashlib
 import json
@@ -47,14 +46,12 @@ from khoj.database.adapters import (
     EntryAdapters,
     FileObjectAdapters,
     aget_user_by_email,
-    ais_user_subscribed,
     create_khoj_token,
     get_default_search_model,
     get_khoj_tokens,
     get_user_name,
     get_user_notion_config,
     get_user_subscription_state,
-    is_user_subscribed,
     run_with_process_lock,
 )
 from khoj.database.models import (
@@ -89,10 +86,6 @@ from khoj.processor.conversation.google.gemini_chat import (
     converse_gemini,
     gemini_send_message_to_model,
 )
-from khoj.processor.conversation.offline.chat_model import (
-    converse_offline,
-    send_message_to_model_offline,
-)
 from khoj.processor.conversation.openai.gpt import (
     converse_openai,
     send_message_to_model,
@@ -117,7 +110,6 @@ from khoj.search_filter.file_filter import FileFilter
 from khoj.search_filter.word_filter import WordFilter
 from khoj.search_type import text_search
 from khoj.utils import state
-from khoj.utils.config import OfflineChatProcessorModel
 from khoj.utils.helpers import (
     LRU,
     ConversationCommand,
@@ -165,17 +157,9 @@ def validate_chat_model(user: KhojUser):
 
 async def is_ready_to_chat(user: KhojUser):
     user_chat_model = await ConversationAdapters.aget_user_chat_model(user)
-    if user_chat_model == None:
+    if user_chat_model is None:
         user_chat_model = await ConversationAdapters.aget_default_chat_model(user)
 
-    if user_chat_model and user_chat_model.model_type == ChatModel.ModelType.OFFLINE:
-        chat_model_name = user_chat_model.name
-        max_tokens = user_chat_model.max_prompt_size
-        if state.offline_chat_processor_config is None:
-            logger.info("Loading Offline Chat Model...")
-            state.offline_chat_processor_config = OfflineChatProcessorModel(chat_model_name, max_tokens)
-        return True
-
     if (
         user_chat_model
         and (
@@ -231,7 +215,6 @@ def update_telemetry_state(
             telemetry_type=telemetry_type,
             api=api,
             client=client,
-            app_config=state.config.app,
             disable_telemetry_env=state.telemetry_disabled,
             properties=user_state,
         )
@@ -595,7 +578,7 @@ async def generate_online_subqueries(
             )
             return {q}
         return response
-    except Exception as e:
+    except Exception:
         logger.error(f"Invalid response for constructing online subqueries: {response}. Returning original query: {q}")
         return {q}
 
@@ -1186,8 +1169,8 @@ async def search_documents(
         agent_has_entries = await sync_to_async(EntryAdapters.agent_has_entries)(agent=agent)
 
     if (
-        not ConversationCommand.Notes in conversation_commands
-        and not ConversationCommand.Default in conversation_commands
+        ConversationCommand.Notes not in conversation_commands
+        and ConversationCommand.Default not in conversation_commands
         and not agent_has_entries
     ):
         yield compiled_references, inferred_queries, q
@@ -1281,6 +1264,7 @@ async def extract_questions(
     location_data: LocationData = None,
     query_images: Optional[List[str]] = None,
     query_files: str = None,
+    max_queries: int = 5,
     tracer: dict = {},
 ):
     """
@@ -1310,14 +1294,20 @@ async def extract_questions(
         location=location,
         username=username,
         personality_context=personality_context,
+        max_queries=max_queries,
     )
 
     prompt = prompts.extract_questions_user_message.format(text=query, chat_history=chat_history_str)
 
     class DocumentQueries(BaseModel):
-        """Choose searches to run on user documents."""
+        """Choose semantic search queries to run on user documents."""
 
-        queries: List[str] = Field(..., min_items=1, description="List of search queries to run on user documents.")
+        queries: List[str] = Field(
+            ...,
+            min_length=1,
+            max_length=max_queries,
+            description="List of semantic search queries to run on user documents.",
+        )
 
     raw_response = await send_message_to_model_wrapper(
         system_message=system_prompt,
@@ -1339,8 +1329,8 @@ async def extract_questions(
             logger.error(f"Invalid response for constructing subqueries: {response}")
             return [query]
         return queries
-    except:
-        logger.warning(f"LLM returned invalid JSON. Falling back to using user message as search query.")
+    except Exception:
+        logger.warning("LLM returned invalid JSON. Falling back to using user message as search query.")
         return [query]
 
 
@@ -1365,7 +1355,7 @@ async def execute_search(
         return results
 
     if q is None or q == "":
-        logger.warning(f"No query param (q) passed in API call to initiate search")
+        logger.warning("No query param (q) passed in API call to initiate search")
         return results
 
     # initialize variables
@@ -1378,7 +1368,7 @@ async def execute_search(
     if user:
         query_cache_key = f"{user_query}-{n}-{t}-{r}-{max_distance}-{dedupe}"
         if query_cache_key in state.query_cache[user.uuid]:
-            logger.debug(f"Return response from query cache")
+            logger.debug("Return response from query cache")
             return state.query_cache[user.uuid][query_cache_key]
 
     # Encode query with filter terms removed
@@ -1470,12 +1460,6 @@ async def send_message_to_model_wrapper(
     vision_available = chat_model.vision_enabled
     api_key = chat_model.ai_model_api.api_key
     api_base_url = chat_model.ai_model_api.api_base_url
-    loaded_model = None
-
-    if model_type == ChatModel.ModelType.OFFLINE:
-        if state.offline_chat_processor_config is None or state.offline_chat_processor_config.loaded_model is None:
-            state.offline_chat_processor_config = OfflineChatProcessorModel(chat_model_name, max_tokens)
-        loaded_model = state.offline_chat_processor_config.loaded_model
 
     truncated_messages = generate_chatml_messages_with_context(
         user_message=query,
@@ -1483,7 +1467,6 @@ async def send_message_to_model_wrapper(
         system_message=system_message,
         chat_history=chat_history,
         model_name=chat_model_name,
-        loaded_model=loaded_model,
         tokenizer_name=tokenizer,
         max_prompt_size=max_tokens,
         vision_enabled=vision_available,
@@ -1492,18 +1475,7 @@ async def send_message_to_model_wrapper(
         query_files=query_files,
     )
 
-    if model_type == ChatModel.ModelType.OFFLINE:
-        return send_message_to_model_offline(
-            messages=truncated_messages,
-            loaded_model=loaded_model,
-            model_name=chat_model_name,
-            max_prompt_size=max_tokens,
-            streaming=False,
-            response_type=response_type,
-            tracer=tracer,
-        )
-
-    elif model_type == ChatModel.ModelType.OPENAI:
+    if model_type == ChatModel.ModelType.OPENAI:
         return send_message_to_model(
             messages=truncated_messages,
             api_key=api_key,
@@ -1565,19 +1537,12 @@ def send_message_to_model_wrapper_sync(
     vision_available = chat_model.vision_enabled
     api_key = chat_model.ai_model_api.api_key
     api_base_url = chat_model.ai_model_api.api_base_url
-    loaded_model = None
-
-    if model_type == ChatModel.ModelType.OFFLINE:
-        if state.offline_chat_processor_config is None or state.offline_chat_processor_config.loaded_model is None:
-            state.offline_chat_processor_config = OfflineChatProcessorModel(chat_model_name, max_tokens)
-        loaded_model = state.offline_chat_processor_config.loaded_model
 
     truncated_messages = generate_chatml_messages_with_context(
         user_message=message,
         system_message=system_message,
         chat_history=chat_history,
         model_name=chat_model_name,
-        loaded_model=loaded_model,
         max_prompt_size=max_tokens,
         vision_enabled=vision_available,
         model_type=model_type,
@@ -1585,18 +1550,7 @@ def send_message_to_model_wrapper_sync(
         query_files=query_files,
     )
 
-    if model_type == ChatModel.ModelType.OFFLINE:
-        return send_message_to_model_offline(
-            messages=truncated_messages,
-            loaded_model=loaded_model,
-            model_name=chat_model_name,
-            max_prompt_size=max_tokens,
-            streaming=False,
-            response_type=response_type,
-            tracer=tracer,
-        )
-
-    elif model_type == ChatModel.ModelType.OPENAI:
+    if model_type == ChatModel.ModelType.OPENAI:
         return send_message_to_model(
             messages=truncated_messages,
             api_key=api_key,
@@ -1678,30 +1632,7 @@ async def agenerate_chat_response(
                 chat_model = vision_enabled_config
                 vision_available = True
 
-        if chat_model.model_type == "offline":
-            loaded_model = state.offline_chat_processor_config.loaded_model
-            chat_response_generator = converse_offline(
-                # Query
-                user_query=query_to_run,
-                # Context
-                references=compiled_references,
-                online_results=online_results,
-                generated_files=raw_generated_files,
-                generated_asset_results=generated_asset_results,
-                location_data=location_data,
-                user_name=user_name,
-                query_files=query_files,
-                chat_history=chat_history,
-                # Model
-                loaded_model=loaded_model,
-                model_name=chat_model.name,
-                max_prompt_size=chat_model.max_prompt_size,
-                tokenizer_name=chat_model.tokenizer,
-                agent=agent,
-                tracer=tracer,
-            )
-
-        elif chat_model.model_type == ChatModel.ModelType.OPENAI:
+        if chat_model.model_type == ChatModel.ModelType.OPENAI:
             openai_chat_config = chat_model.ai_model_api
             api_key = openai_chat_config.api_key
             chat_model_name = chat_model.name
@@ -1948,8 +1879,8 @@ class ApiUserRateLimiter:
 
         user: KhojUser = websocket.scope["user"].object
         subscribed = has_required_scope(websocket, ["premium"])
-        current_window = "today" if self.window == 60 * 60 * 24 else f"now"
-        next_window = "tomorrow" if self.window == 60 * 60 * 24 else f"in a bit"
+        current_window = "today" if self.window == 60 * 60 * 24 else "now"
+        next_window = "tomorrow" if self.window == 60 * 60 * 24 else "in a bit"
         common_message_prefix = f"I'm glad you're enjoying interacting with me! You've unfortunately exceeded your usage limit for {current_window}."
 
         # Remove requests outside of the time window
@@ -2292,7 +2223,7 @@ def should_notify(original_query: str, executed_query: str, ai_response: str, us
             should_notify_result = response["decision"] == "Yes"
             reason = response.get("reason", "unknown")
             logger.info(
-                f'Decided to {"not " if not should_notify_result else ""}notify user of automation response because of reason: {reason}.'
+                f"Decided to {'not ' if not should_notify_result else ''}notify user of automation response because of reason: {reason}."
             )
             return should_notify_result
         except Exception as e:
@@ -2386,7 +2317,7 @@ def scheduled_chat(
         response_map = raw_response.json()
         ai_response = response_map.get("response") or response_map.get("image")
         is_image = False
-        if type(ai_response) == dict:
+        if isinstance(ai_response, dict):
             is_image = ai_response.get("image") is not None
     else:
         ai_response = raw_response.text
@@ -2533,12 +2464,12 @@ async def aschedule_automation(
 
 def construct_automation_created_message(automation: Job, crontime: str, query_to_run: str, subject: str):
     # Display next run time in user timezone instead of UTC
-    schedule = f'{cron_descriptor.get_description(crontime)} {automation.next_run_time.strftime("%Z")}'
+    schedule = f"{cron_descriptor.get_description(crontime)} {automation.next_run_time.strftime('%Z')}"
     next_run_time = automation.next_run_time.strftime("%Y-%m-%d %I:%M %p %Z")
     # Remove /automated_task prefix from inferred_query
     unprefixed_query_to_run = re.sub(r"^\/automated_task\s*", "", query_to_run)
     # Create the automation response
-    automation_icon_url = f"/static/assets/icons/automation.svg"
+    automation_icon_url = "/static/assets/icons/automation.svg"
     return f"""
     ### ![]({automation_icon_url}) Created Automation
 - Subject: **{subject}**
@@ -2786,19 +2717,20 @@ def configure_content(
     t: Optional[state.SearchType] = state.SearchType.All,
 ) -> bool:
     success = True
-    if t == None:
+    if t is None:
         t = state.SearchType.All
 
     if t is not None and t in [type.value for type in state.SearchType]:
         t = state.SearchType(t)
 
-    if t is not None and not t.value in [type.value for type in state.SearchType]:
+    if t is not None and t.value not in [type.value for type in state.SearchType]:
         logger.warning(f"🚨 Invalid search type: {t}")
         return False
 
     search_type = t.value if t else None
 
-    no_documents = all([not files.get(file_type) for file_type in files])
+    # Check if client sent any documents of the supported types
+    no_client_sent_documents = all([not files.get(file_type) for file_type in files])
 
     if files is None:
         logger.warning(f"🚨 No files to process for {search_type} search.")
@@ -2872,7 +2804,8 @@ def configure_content(
         success = False
 
     try:
-        if no_documents:
+        # Run server side indexing of user Github docs if no client sent documents
+        if no_client_sent_documents:
             github_config = GithubConfig.objects.filter(user=user).prefetch_related("githubrepoconfig").first()
             if (
                 search_type == state.SearchType.All.value or search_type == state.SearchType.Github.value
@@ -2892,7 +2825,8 @@ def configure_content(
         success = False
 
     try:
-        if no_documents:
+        # Run server side indexing of user Notion docs if no client sent documents
+        if no_client_sent_documents:
             # Initialize Notion Search
             notion_config = NotionConfig.objects.filter(user=user).first()
             if (
@@ -3058,7 +2992,7 @@ async def grep_files(
             query += f" {' and '.join(context_info)}"
         if line_count > max_results:
             if lines_before or lines_after:
-                query += f" for"
+                query += " for"
             query += f" first {max_results} results"
         return query
 
@@ -3068,7 +3002,7 @@ async def grep_files(
     lines_after = lines_after or 0
 
     try:
-        regex = re.compile(regex_pattern, re.IGNORECASE)
+        regex = re.compile(regex_pattern, re.IGNORECASE | re.MULTILINE)
     except re.error as e:
         yield {
             "query": _generate_query(0, 0, path_prefix, regex_pattern, lines_before, lines_after),
@@ -3078,7 +3012,14 @@ async def grep_files(
         return
 
     try:
-        file_matches = await FileObjectAdapters.aget_file_objects_by_regex(user, regex_pattern, path_prefix)
+        # Make db pushdown filters more permissive by removing line anchors
+        # The precise line-anchored matching will be done in Python stage
+        db_pattern = regex_pattern
+        db_pattern = re.sub(r"\(\?\w*\)", "", db_pattern)  # Remove inline flags like (?i), (?m), (?im)
+        db_pattern = re.sub(r"^\^", "", db_pattern)  # Remove ^ at regex pattern start
+        db_pattern = re.sub(r"\$$", "", db_pattern)  # Remove $ at regex pattern end
+
+        file_matches = await FileObjectAdapters.aget_file_objects_by_regex(user, db_pattern, path_prefix)
 
         line_matches = []
         for file_object in file_matches:

khoj/routers/research.py

@@ -15,7 +15,6 @@ from khoj.processor.conversation.utils import (
     ResearchIteration,
     ToolCall,
     construct_iteration_history,
-    construct_structured_message,
     construct_tool_chat_history,
     load_complex_json,
 )
@@ -24,7 +23,6 @@ from khoj.processor.tools.online_search import read_webpages_content, search_onl
 from khoj.processor.tools.run_code import run_code
 from khoj.routers.helpers import (
     ChatEvent,
-    generate_summary_from_files,
     get_message_from_queue,
     grep_files,
     list_files,
@@ -102,7 +100,7 @@ async def apick_next_tool(
         ConversationCommand.Notes.value: [tool.value for tool in document_research_tools],
         ConversationCommand.Webpage.value: [ConversationCommand.ReadWebpage.value],
         ConversationCommand.Online.value: [ConversationCommand.SearchWeb.value],
-        ConversationCommand.Code.value: [ConversationCommand.RunCode.value],
+        ConversationCommand.Code.value: [ConversationCommand.PythonCoder.value],
         ConversationCommand.Operator.value: [ConversationCommand.OperateComputer.value],
     }
     for input_tool, research_tools in input_tools_to_research_tools.items():
@@ -184,7 +182,7 @@ async def apick_next_tool(
         # TODO: Handle multiple tool calls.
         response_text = response.text
         parsed_response = [ToolCall(**item) for item in load_complex_json(response_text)][0]
-    except Exception as e:
+    except Exception:
         # Otherwise assume the model has decided to end the research run and respond to the user.
         parsed_response = ToolCall(name=ConversationCommand.Text, args={"response": response_text}, id=None)
 
@@ -199,7 +197,7 @@ async def apick_next_tool(
         if i.warning is None and isinstance(i.query, ToolCall)
     }
     if (parsed_response.name, dict_to_tuple(parsed_response.args)) in previous_tool_query_combinations:
-        warning = f"Repeated tool, query combination detected. Skipping iteration. Try something different."
+        warning = "Repeated tool, query combination detected. Skipping iteration. Try something different."
     # Only send client status updates if we'll execute this iteration and model has thoughts to share.
     elif send_status_func and not is_none_or_empty(response.thought):
         async for event in send_status_func(response.thought):
@@ -414,11 +412,13 @@ async def research(
                 this_iteration.warning = f"Error reading webpages: {e}"
                 logger.error(this_iteration.warning, exc_info=True)
 
-        elif this_iteration.query.name == ConversationCommand.RunCode:
+        elif this_iteration.query.name == ConversationCommand.PythonCoder:
             try:
                 async for result in run_code(
                     **this_iteration.query.args,
-                    conversation_history=construct_tool_chat_history(previous_iterations, ConversationCommand.RunCode),
+                    conversation_history=construct_tool_chat_history(
+                        previous_iterations, ConversationCommand.PythonCoder
+                    ),
                     context="",
                     location_data=location,
                     user=user,
@@ -435,7 +435,7 @@ async def research(
                         this_iteration.codeContext = code_results
                 async for result in send_status_func(f"**Ran code snippets**: {len(this_iteration.codeContext)}"):
                     yield result
-            except ValueError as e:
+            except (ValueError, TypeError) as e:
                 this_iteration.warning = f"Error running code: {e}"
                 logger.warning(this_iteration.warning, exc_info=True)
 

khoj/search_filter/base_filter.py

@@ -4,12 +4,10 @@ from typing import List
 
 class BaseFilter(ABC):
     @abstractmethod
-    def get_filter_terms(self, query: str) -> List[str]:
-        ...
+    def get_filter_terms(self, query: str) -> List[str]: ...
 
     def can_filter(self, raw_query: str) -> bool:
         return len(self.get_filter_terms(raw_query)) > 0
 
     @abstractmethod
-    def defilter(self, query: str) -> str:
-        ...
+    def defilter(self, query: str) -> str: ...

khoj/search_type/text_search.py

@@ -9,9 +9,8 @@ from asgiref.sync import sync_to_async
 from sentence_transformers import util
 
 from khoj.database.adapters import EntryAdapters, get_default_search_model
-from khoj.database.models import Agent
+from khoj.database.models import Agent, KhojUser
 from khoj.database.models import Entry as DbEntry
-from khoj.database.models import KhojUser
 from khoj.processor.content.text_to_entries import TextToEntries
 from khoj.utils import state
 from khoj.utils.helpers import get_absolute_path, timer

khoj/utils/cli.py

@@ -1,36 +1,19 @@
 import argparse
 import logging
-import os
 import pathlib
 from importlib.metadata import version
 
 logger = logging.getLogger(__name__)
 
-from khoj.migrations.migrate_offline_chat_default_model import (
-    migrate_offline_chat_default_model,
-)
-from khoj.migrations.migrate_offline_chat_schema import migrate_offline_chat_schema
-from khoj.migrations.migrate_offline_model import migrate_offline_model
-from khoj.migrations.migrate_processor_config_openai import (
-    migrate_processor_conversation_schema,
-)
-from khoj.migrations.migrate_server_pg import migrate_server_pg
-from khoj.migrations.migrate_version import migrate_config_to_version
-from khoj.utils.helpers import is_env_var_true, resolve_absolute_path
-from khoj.utils.yaml import parse_config_from_file
-
 
 def cli(args=None):
     # Setup Argument Parser for the Commandline Interface
     parser = argparse.ArgumentParser(description="Start Khoj; An AI personal assistant for your Digital Brain")
     parser.add_argument(
-        "--config-file", default="~/.khoj/khoj.yml", type=pathlib.Path, help="YAML file to configure Khoj"
-    )
-    parser.add_argument(
-        "--regenerate",
-        action="store_true",
-        default=False,
-        help="Regenerate model embeddings from source files. Default: false",
+        "--log-file",
+        default="~/.khoj/khoj.log",
+        type=pathlib.Path,
+        help="File path for server logs. Default: ~/.khoj/khoj.log",
     )
     parser.add_argument("--verbose", "-v", action="count", default=0, help="Show verbose conversion logs. Default: 0")
     parser.add_argument("--host", type=str, default="127.0.0.1", help="Host address of the server. Default: 127.0.0.1")
@@ -43,14 +26,11 @@ def cli(args=None):
     parser.add_argument("--sslcert", type=str, help="Path to SSL certificate file")
     parser.add_argument("--sslkey", type=str, help="Path to SSL key file")
     parser.add_argument("--version", "-V", action="store_true", help="Print the installed Khoj version and exit")
-    parser.add_argument(
-        "--disable-chat-on-gpu", action="store_true", default=False, help="Disable using GPU for the offline chat model"
-    )
     parser.add_argument(
         "--anonymous-mode",
         action="store_true",
         default=False,
-        help="Run Khoj in anonymous mode. This does not require any login for connecting users.",
+        help="Run Khoj in single user mode with no login required. Useful for personal use or testing.",
     )
     parser.add_argument(
         "--non-interactive",
@@ -64,38 +44,10 @@ def cli(args=None):
     if len(remaining_args) > 0:
         logger.info(f"⚠️  Ignoring unknown commandline args: {remaining_args}")
 
-    # Set default values for arguments
-    args.chat_on_gpu = not args.disable_chat_on_gpu
-
     args.version_no = version("khoj")
     if args.version:
         # Show version of khoj installed and exit
         print(args.version_no)
         exit(0)
 
-    # Normalize config_file path to absolute path
-    args.config_file = resolve_absolute_path(args.config_file)
-
-    if not args.config_file.exists():
-        args.config = None
-    else:
-        args = run_migrations(args)
-        args.config = parse_config_from_file(args.config_file)
-        if is_env_var_true("KHOJ_TELEMETRY_DISABLE"):
-            args.config.app.should_log_telemetry = False
-
-    return args
-
-
-def run_migrations(args):
-    migrations = [
-        migrate_config_to_version,
-        migrate_processor_conversation_schema,
-        migrate_offline_model,
-        migrate_offline_chat_schema,
-        migrate_offline_chat_default_model,
-        migrate_server_pg,
-    ]
-    for migration in migrations:
-        args = migration(args)
     return args

khoj/utils/config.py

@@ -1,22 +1,7 @@
 # System Packages
 from __future__ import annotations  # to avoid quoting type hints
 
-import logging
-from dataclasses import dataclass
 from enum import Enum
-from typing import TYPE_CHECKING, Any, List, Optional, Union
-
-import torch
-
-from khoj.processor.conversation.offline.utils import download_model
-
-logger = logging.getLogger(__name__)
-
-
-if TYPE_CHECKING:
-    from sentence_transformers import CrossEncoder
-
-    from khoj.utils.models import BaseEncoder
 
 
 class SearchType(str, Enum):
@@ -29,53 +14,3 @@ class SearchType(str, Enum):
     Notion = "notion"
     Plaintext = "plaintext"
     Docx = "docx"
-
-
-class ProcessorType(str, Enum):
-    Conversation = "conversation"
-
-
-@dataclass
-class TextContent:
-    enabled: bool
-
-
-@dataclass
-class ImageContent:
-    image_names: List[str]
-    image_embeddings: torch.Tensor
-    image_metadata_embeddings: torch.Tensor
-
-
-@dataclass
-class TextSearchModel:
-    bi_encoder: BaseEncoder
-    cross_encoder: Optional[CrossEncoder] = None
-    top_k: Optional[int] = 15
-
-
-@dataclass
-class ImageSearchModel:
-    image_encoder: BaseEncoder
-
-
-@dataclass
-class SearchModels:
-    text_search: Optional[TextSearchModel] = None
-
-
-@dataclass
-class OfflineChatProcessorConfig:
-    loaded_model: Union[Any, None] = None
-
-
-class OfflineChatProcessorModel:
-    def __init__(self, chat_model: str = "bartowski/Meta-Llama-3.1-8B-Instruct-GGUF", max_tokens: int = None):
-        self.chat_model = chat_model
-        self.loaded_model = None
-        try:
-            self.loaded_model = download_model(self.chat_model, max_tokens=max_tokens)
-        except ValueError as e:
-            self.loaded_model = None
-            logger.error(f"Error while loading offline chat model: {e}", exc_info=True)
-            raise e

khoj/utils/constants.py

@@ -10,13 +10,6 @@ empty_escape_sequences = "\n|\r|\t| "
 app_env_filepath = "~/.khoj/env"
 telemetry_server = "https://khoj.beta.haletic.com/v1/telemetry"
 content_directory = "~/.khoj/content/"
-default_offline_chat_models = [
-    "bartowski/Meta-Llama-3.1-8B-Instruct-GGUF",
-    "bartowski/Llama-3.2-3B-Instruct-GGUF",
-    "bartowski/gemma-2-9b-it-GGUF",
-    "bartowski/gemma-2-2b-it-GGUF",
-    "bartowski/Qwen2.5-14B-Instruct-GGUF",
-]
 default_openai_chat_models = ["gpt-4o-mini", "gpt-4.1", "o3", "o4-mini"]
 default_gemini_chat_models = ["gemini-2.0-flash", "gemini-2.5-flash-preview-05-20", "gemini-2.5-pro-preview-06-05"]
 default_anthropic_chat_models = ["claude-sonnet-4-0", "claude-3-5-haiku-latest"]
@@ -47,6 +40,9 @@ model_to_cost: Dict[str, Dict[str, float]] = {
     "o3": {"input": 2.0, "output": 8.00},
     "o3-pro": {"input": 20.0, "output": 80.00},
     "o4-mini": {"input": 1.10, "output": 4.40},
+    "gpt-5-2025-08-07": {"input": 1.25, "output": 10.00, "cache_read": 0.125},
+    "gpt-5-mini-2025-08-07": {"input": 0.25, "output": 2.00, "cache_read": 0.025},
+    "gpt-5-nano-2025-08-07": {"input": 0.05, "output": 0.40, "cache_read": 0.005},
     # 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},
@@ -75,4 +71,7 @@ model_to_cost: Dict[str, Dict[str, float]] = {
     "grok-3-latest": {"input": 3.0, "output": 15.0},
     "grok-3-mini": {"input": 0.30, "output": 0.50},
     "grok-3-mini-latest": {"input": 0.30, "output": 0.50},
+    # Groq pricing
+    "moonshotai/kimi-k2-instruct": {"input": 1.00, "output": 3.00},
+    "openai/gpt-oss-120b": {"input": 0.15, "output": 0.75},
 }