improve scroll

* trying out a fix

* add ability to manually run model tests

---------

Co-authored-by: Richard Kuo (Danswer) <rkuo@onyx.app>

.github/workflows/pr-integration-tests.yml

@@ -145,7 +145,7 @@ jobs:
         run: |
           cd deployment/docker_compose
           docker compose -f docker-compose.multitenant-dev.yml -p onyx-stack down -v
-      
+
       # NOTE: Use pre-ping/null pool to reduce flakiness due to dropped connections
       - name: Start Docker containers
         run: |
@@ -157,6 +157,7 @@ jobs:
           REQUIRE_EMAIL_VERIFICATION=false \
           DISABLE_TELEMETRY=true \
           IMAGE_TAG=test \
+          INTEGRATION_TESTS_MODE=true \
           docker compose -f docker-compose.dev.yml -p onyx-stack up -d
         id: start_docker
 
@@ -199,7 +200,7 @@ jobs:
           cd backend/tests/integration/mock_services
           docker compose -f docker-compose.mock-it-services.yml \
             -p mock-it-services-stack up -d
-      
+
       # NOTE: Use pre-ping/null to reduce flakiness due to dropped connections
       - name: Run Standard Integration Tests
         run: |

.github/workflows/pr-python-model-tests.yml

@@ -1,10 +1,16 @@
-name: Connector Tests
+name: Model Server Tests
 
 on:
   schedule:
     # This cron expression runs the job daily at 16:00 UTC (9am PT)
     - cron: "0 16 * * *"
-
+  workflow_dispatch:
+    inputs:
+      branch:
+        description: 'Branch to run the workflow on'
+        required: false
+        default: 'main'
+        
 env:
   # Bedrock
   AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
@@ -26,6 +32,23 @@ jobs:
       - name: Checkout code
         uses: actions/checkout@v4
 
+      - name: Login to Docker Hub
+        uses: docker/login-action@v3
+        with:
+          username: ${{ secrets.DOCKER_USERNAME }}
+          password: ${{ secrets.DOCKER_TOKEN }}
+
+      # tag every docker image with "test" so that we can spin up the correct set
+      # of images during testing
+
+      # We don't need to build the Web Docker image since it's not yet used
+      # in the integration tests. We have a separate action to verify that it builds
+      # successfully.
+      - name: Pull Model Server Docker image
+        run: |
+          docker pull onyxdotapp/onyx-model-server:latest
+          docker tag onyxdotapp/onyx-model-server:latest onyxdotapp/onyx-model-server:test
+          
       - name: Set up Python
         uses: actions/setup-python@v5
         with:
@@ -41,6 +64,49 @@ jobs:
           pip install --retries 5 --timeout 30 -r backend/requirements/default.txt
           pip install --retries 5 --timeout 30 -r backend/requirements/dev.txt
 
+      - name: Start Docker containers
+        run: |
+          cd deployment/docker_compose
+          ENABLE_PAID_ENTERPRISE_EDITION_FEATURES=true \
+          AUTH_TYPE=basic \
+          REQUIRE_EMAIL_VERIFICATION=false \
+          DISABLE_TELEMETRY=true \
+          IMAGE_TAG=test \
+          docker compose -f docker-compose.dev.yml -p onyx-stack up -d indexing_model_server
+        id: start_docker
+
+      - name: Wait for service to be ready
+        run: |
+          echo "Starting wait-for-service script..."
+
+          start_time=$(date +%s)
+          timeout=300  # 5 minutes in seconds
+
+          while true; do
+            current_time=$(date +%s)
+            elapsed_time=$((current_time - start_time))
+            
+            if [ $elapsed_time -ge $timeout ]; then
+              echo "Timeout reached. Service did not become ready in 5 minutes."
+              exit 1
+            fi
+            
+            # Use curl with error handling to ignore specific exit code 56
+            response=$(curl -s -o /dev/null -w "%{http_code}" http://localhost:9000/api/health || echo "curl_error")
+            
+            if [ "$response" = "200" ]; then
+              echo "Service is ready!"
+              break
+            elif [ "$response" = "curl_error" ]; then
+              echo "Curl encountered an error, possibly exit code 56. Continuing to retry..."
+            else
+              echo "Service not ready yet (HTTP status $response). Retrying in 5 seconds..."
+            fi
+            
+            sleep 5
+          done
+          echo "Finished waiting for service."
+          
       - name: Run Tests
         shell: script -q -e -c "bash --noprofile --norc -eo pipefail {0}"
         run: |
@@ -56,3 +122,10 @@ jobs:
             -H 'Content-type: application/json' \
             --data '{"text":"Scheduled Model Tests failed! Check the run at: https://github.com/${{ github.repository }}/actions/runs/${{ github.run_id }}"}' \
             $SLACK_WEBHOOK
+            
+      - name: Stop Docker containers
+        if: always()
+        run: |
+          cd deployment/docker_compose
+          docker compose -f docker-compose.dev.yml -p onyx-stack down -v
+          

README.md

@@ -26,12 +26,12 @@
 
 <strong>[Onyx](https://www.onyx.app/)</strong> (formerly Danswer) is the AI platform connected to your company's docs, apps, and people.
 Onyx provides a feature rich Chat interface and plugs into any LLM of your choice.
-There are over 40 supported connectors such as Google Drive, Slack, Confluence, Salesforce, etc. which keep knowledge and permissions up to date.
-Create custom AI agents with unique prompts, knowledge, and actions the agents can take.
+Keep knowledge and access controls sync-ed across over 40 connectors like Google Drive, Slack, Confluence, Salesforce, etc.
+Create custom AI agents with unique prompts, knowledge, and actions that the agents can take.
 Onyx can be deployed securely anywhere and for any scale - on a laptop, on-premise, or to cloud.
 
 
-<h3>Feature Showcase</h3>
+<h3>Feature Highlights</h3>
 
 **Deep research over your team's knowledge:**
 
@@ -63,22 +63,21 @@ We also have built-in support for high-availability/scalable deployment on Kuber
 References [here](https://github.com/onyx-dot-app/onyx/tree/main/deployment).
 
 
+## 🔍 Other Notable Benefits of Onyx
+- Custom deep learning models for indexing and inference time, only through Onyx + learning from user feedback.
+- Flexible security features like SSO (OIDC/SAML/OAuth2), RBAC, encryption of credentials, etc.
+- Knowledge curation features like document-sets, query history, usage analytics, etc.
+- Scalable deployment options tested up to many tens of thousands users and hundreds of millions of documents.
+
+
 ## 🚧 Roadmap
-- Extensions to the Chrome Plugin
-- Latest methods in information retrieval (StructRAG, LightGraphRAG, etc.)
+- New methods in information retrieval (StructRAG, LightGraphRAG, etc.)
 - Personalized Search
 - Organizational understanding and ability to locate and suggest experts from your team.
 - Code Search
 - SQL and Structured Query Language
 
 
-## 🔍 Other Notable Benefits of Onyx
-- Custom deep learning models only through Onyx + learn from user feedback.
-- Flexible security features like SSO (OIDC/SAML/OAuth2), RBAC, encryption of credentials, etc.
-- Knowledge curation features like document-sets, query history, usage analytics, etc.
-- Scalable deployment options tested up to many tens of thousands users and hundreds of millions of documents.
-
-
 ## 🔌 Connectors
 Keep knowledge and access up to sync across 40+ connectors:
 

backend/model_server/encoders.py

@@ -98,12 +98,17 @@ class CloudEmbedding:
             return final_embeddings
         except Exception as e:
             error_string = (
-                f"Error embedding text with OpenAI: {str(e)} \n"
-                f"Model: {model} \n"
-                f"Provider: {self.provider} \n"
-                f"Texts: {texts}"
+                f"Exception embedding text with OpenAI - {type(e)}: "
+                f"Model: {model} "
+                f"Provider: {self.provider} "
+                f"Exception: {e}"
             )
             logger.error(error_string)
+
+            # only log text when it's not an authentication error.
+            if not isinstance(e, openai.AuthenticationError):
+                logger.debug(f"Exception texts: {texts}")
+
             raise RuntimeError(error_string)
 
     async def _embed_cohere(

backend/onyx/background/celery/tasks/external_group_syncing/tasks.py

@@ -361,6 +361,7 @@ def connector_external_group_sync_generator_task(
             cc_pair = get_connector_credential_pair_from_id(
                 db_session=db_session,
                 cc_pair_id=cc_pair_id,
+                eager_load_credential=True,
             )
             if cc_pair is None:
                 raise ValueError(

backend/onyx/background/indexing/run_indexing.py

@@ -15,6 +15,7 @@ from onyx.background.indexing.memory_tracer import MemoryTracer
 from onyx.configs.app_configs import INDEX_BATCH_SIZE
 from onyx.configs.app_configs import INDEXING_SIZE_WARNING_THRESHOLD
 from onyx.configs.app_configs import INDEXING_TRACER_INTERVAL
+from onyx.configs.app_configs import INTEGRATION_TESTS_MODE
 from onyx.configs.app_configs import LEAVE_CONNECTOR_ACTIVE_ON_INITIALIZATION_FAILURE
 from onyx.configs.app_configs import POLL_CONNECTOR_OFFSET
 from onyx.configs.constants import DocumentSource
@@ -89,8 +90,8 @@ def _get_connector_runner(
         )
 
         # validate the connector settings
-
-        runnable_connector.validate_connector_settings()
+        if not INTEGRATION_TESTS_MODE:
+            runnable_connector.validate_connector_settings()
 
     except Exception as e:
         logger.exception(f"Unable to instantiate connector due to {e}")

backend/onyx/configs/app_configs.py

@@ -158,7 +158,7 @@ POSTGRES_USER = os.environ.get("POSTGRES_USER") or "postgres"
 POSTGRES_PASSWORD = urllib.parse.quote_plus(
     os.environ.get("POSTGRES_PASSWORD") or "password"
 )
-POSTGRES_HOST = os.environ.get("POSTGRES_HOST") or "localhost"
+POSTGRES_HOST = os.environ.get("POSTGRES_HOST") or "127.0.0.1"
 POSTGRES_PORT = os.environ.get("POSTGRES_PORT") or "5432"
 POSTGRES_DB = os.environ.get("POSTGRES_DB") or "postgres"
 AWS_REGION_NAME = os.environ.get("AWS_REGION_NAME") or "us-east-2"
@@ -626,6 +626,8 @@ POD_NAMESPACE = os.environ.get("POD_NAMESPACE")
 
 DEV_MODE = os.environ.get("DEV_MODE", "").lower() == "true"
 
+INTEGRATION_TESTS_MODE = os.environ.get("INTEGRATION_TESTS_MODE", "").lower() == "true"
+
 MOCK_CONNECTOR_FILE_PATH = os.environ.get("MOCK_CONNECTOR_FILE_PATH")
 
 TEST_ENV = os.environ.get("TEST_ENV", "").lower() == "true"

backend/onyx/connectors/factory.py

@@ -3,6 +3,7 @@ from typing import Type
 
 from sqlalchemy.orm import Session
 
+from onyx.configs.app_configs import INTEGRATION_TESTS_MODE
 from onyx.configs.constants import DocumentSource
 from onyx.configs.constants import DocumentSourceRequiringTenantContext
 from onyx.connectors.airtable.airtable_connector import AirtableConnector
@@ -187,6 +188,9 @@ def validate_ccpair_for_user(
     user: User | None,
     tenant_id: str | None,
 ) -> None:
+    if INTEGRATION_TESTS_MODE:
+        return
+
     # Validate the connector settings
     connector = fetch_connector_by_id(connector_id, db_session)
     credential = fetch_credential_by_id_for_user(
@@ -199,7 +203,10 @@ def validate_ccpair_for_user(
     if not connector:
         raise ValueError("Connector not found")
 
-    if connector.source == DocumentSource.INGESTION_API:
+    if (
+        connector.source == DocumentSource.INGESTION_API
+        or connector.source == DocumentSource.MOCK_CONNECTOR
+    ):
         return
 
     if not credential:

backend/onyx/connectors/google_drive/connector.py

@@ -220,7 +220,14 @@ class GoogleDriveConnector(LoadConnector, PollConnector, SlimConnector):
         return self._creds
 
     def load_credentials(self, credentials: dict[str, Any]) -> dict[str, str] | None:
-        self._primary_admin_email = credentials[DB_CREDENTIALS_PRIMARY_ADMIN_KEY]
+        try:
+            self._primary_admin_email = credentials[DB_CREDENTIALS_PRIMARY_ADMIN_KEY]
+        except KeyError:
+            raise ValueError(
+                "Primary admin email missing, "
+                "should not call this property "
+                "before calling load_credentials"
+            )
 
         self._creds, new_creds_dict = get_google_creds(
             credentials=credentials,

backend/onyx/db/connector_credential_pair.py

@@ -194,9 +194,14 @@ def get_connector_credential_pair_from_id_for_user(
 def get_connector_credential_pair_from_id(
     db_session: Session,
     cc_pair_id: int,
+    eager_load_credential: bool = False,
 ) -> ConnectorCredentialPair | None:
     stmt = select(ConnectorCredentialPair).distinct()
     stmt = stmt.where(ConnectorCredentialPair.id == cc_pair_id)
+
+    if eager_load_credential:
+        stmt = stmt.options(joinedload(ConnectorCredentialPair.credential))
+
     result = db_session.execute(stmt)
     return result.scalar_one_or_none()
 

backend/scripts/debugging/onyx_redis.py

@@ -3,6 +3,7 @@ import json
 import logging
 import sys
 import time
+from enum import Enum
 from logging import getLogger
 from typing import cast
 from uuid import UUID
@@ -20,10 +21,13 @@ from onyx.configs.app_configs import REDIS_PORT
 from onyx.configs.app_configs import REDIS_SSL
 from onyx.db.engine import get_session_with_tenant
 from onyx.db.users import get_user_by_email
+from onyx.redis.redis_connector import RedisConnector
+from onyx.redis.redis_connector_index import RedisConnectorIndex
 from onyx.redis.redis_pool import RedisPool
 from shared_configs.configs import MULTI_TENANT
 from shared_configs.configs import POSTGRES_DEFAULT_SCHEMA
 from shared_configs.contextvars import CURRENT_TENANT_ID_CONTEXTVAR
+from shared_configs.contextvars import get_current_tenant_id
 
 # Tool to run helpful operations on Redis in production
 # This is targeted for internal usage and may not have all the necessary parameters
@@ -42,6 +46,19 @@ SCAN_ITER_COUNT = 10000
 BATCH_DEFAULT = 1000
 
 
+class OnyxRedisCommand(Enum):
+    purge_connectorsync_taskset = "purge_connectorsync_taskset"
+    purge_documentset_taskset = "purge_documentset_taskset"
+    purge_usergroup_taskset = "purge_usergroup_taskset"
+    purge_locks_blocking_deletion = "purge_locks_blocking_deletion"
+    purge_vespa_syncing = "purge_vespa_syncing"
+    get_user_token = "get_user_token"
+    delete_user_token = "delete_user_token"
+
+    def __str__(self) -> str:
+        return self.value
+
+
 def get_user_id(user_email: str) -> tuple[UUID, str]:
     tenant_id = (
         get_tenant_id_for_email(user_email) if MULTI_TENANT else POSTGRES_DEFAULT_SCHEMA
@@ -55,50 +72,79 @@ def get_user_id(user_email: str) -> tuple[UUID, str]:
 
 
 def onyx_redis(
-    command: str,
+    command: OnyxRedisCommand,
     batch: int,
     dry_run: bool,
+    ssl: bool,
     host: str,
     port: int,
     db: int,
     password: str | None,
     user_email: str | None = None,
+    cc_pair_id: int | None = None,
 ) -> int:
+    # this is global and not tenant aware
     pool = RedisPool.create_pool(
         host=host,
         port=port,
         db=db,
         password=password if password else "",
-        ssl=REDIS_SSL,
+        ssl=ssl,
         ssl_cert_reqs="optional",
         ssl_ca_certs=None,
     )
 
     r = Redis(connection_pool=pool)
 
+    logger.info("Redis ping starting. This may hang if your settings are incorrect.")
+
     try:
         r.ping()
     except:
         logger.exception("Redis ping exceptioned")
         raise
 
-    if command == "purge_connectorsync_taskset":
+    logger.info("Redis ping succeeded.")
+
+    if command == OnyxRedisCommand.purge_connectorsync_taskset:
         """Purge connector tasksets. Used when the tasks represented in the tasksets
         have been purged."""
         return purge_by_match_and_type(
             "*connectorsync_taskset*", "set", batch, dry_run, r
         )
-    elif command == "purge_documentset_taskset":
+    elif command == OnyxRedisCommand.purge_documentset_taskset:
         return purge_by_match_and_type(
             "*documentset_taskset*", "set", batch, dry_run, r
         )
-    elif command == "purge_usergroup_taskset":
+    elif command == OnyxRedisCommand.purge_usergroup_taskset:
         return purge_by_match_and_type("*usergroup_taskset*", "set", batch, dry_run, r)
-    elif command == "purge_vespa_syncing":
+    elif command == OnyxRedisCommand.purge_locks_blocking_deletion:
+        if cc_pair_id is None:
+            logger.error("You must specify --cc-pair with purge_deletion_locks")
+            return 1
+
+        tenant_id = get_current_tenant_id()
+        logger.info(f"Purging locks associated with deleting cc_pair={cc_pair_id}.")
+        redis_connector = RedisConnector(tenant_id, cc_pair_id)
+
+        match_pattern = f"{tenant_id}:{RedisConnectorIndex.FENCE_PREFIX}_{cc_pair_id}/*"
+        purge_by_match_and_type(match_pattern, "string", batch, dry_run, r)
+
+        redis_delete_if_exists_helper(
+            f"{tenant_id}:{redis_connector.prune.fence_key}", dry_run, r
+        )
+        redis_delete_if_exists_helper(
+            f"{tenant_id}:{redis_connector.permissions.fence_key}", dry_run, r
+        )
+        redis_delete_if_exists_helper(
+            f"{tenant_id}:{redis_connector.external_group_sync.fence_key}", dry_run, r
+        )
+        return 0
+    elif command == OnyxRedisCommand.purge_vespa_syncing:
         return purge_by_match_and_type(
             "*connectorsync:vespa_syncing*", "string", batch, dry_run, r
         )
-    elif command == "get_user_token":
+    elif command == OnyxRedisCommand.get_user_token:
         if not user_email:
             logger.error("You must specify --user-email with get_user_token")
             return 1
@@ -109,7 +155,7 @@ def onyx_redis(
         else:
             print(f"No token found for user {user_email}")
             return 2
-    elif command == "delete_user_token":
+    elif command == OnyxRedisCommand.delete_user_token:
         if not user_email:
             logger.error("You must specify --user-email with delete_user_token")
             return 1
@@ -131,6 +177,25 @@ def flush_batch_delete(batch_keys: list[bytes], r: Redis) -> None:
         pipe.execute()
 
 
+def redis_delete_if_exists_helper(key: str, dry_run: bool, r: Redis) -> bool:
+    """Returns True if the key was found, False if not.
+    This function exists for logging purposes as the delete operation itself
+    doesn't really need to check the existence of the key.
+    """
+
+    if not r.exists(key):
+        logger.info(f"Did not find {key}.")
+        return False
+
+    if dry_run:
+        logger.info(f"(DRY-RUN) Deleting {key}.")
+    else:
+        logger.info(f"Deleting {key}.")
+        r.delete(key)
+
+    return True
+
+
 def purge_by_match_and_type(
     match_pattern: str, match_type: str, batch_size: int, dry_run: bool, r: Redis
 ) -> int:
@@ -138,6 +203,12 @@ def purge_by_match_and_type(
     match_type: https://redis.io/docs/latest/commands/type/
     """
 
+    logger.info(
+        f"purge_by_match_and_type start: "
+        f"match_pattern={match_pattern} "
+        f"match_type={match_type}"
+    )
+
     # cursor = "0"
     # while cursor != 0:
     #     cursor, data = self.scan(
@@ -164,13 +235,15 @@ def purge_by_match_and_type(
         logger.info(f"Deleting item {count}: {key_str}")
 
         batch_keys.append(key)
+
+        # flush if batch size has been reached
         if len(batch_keys) >= batch_size:
             flush_batch_delete(batch_keys, r)
             batch_keys.clear()
 
-    if len(batch_keys) >= batch_size:
-        flush_batch_delete(batch_keys, r)
-        batch_keys.clear()
+    # final flush
+    flush_batch_delete(batch_keys, r)
+    batch_keys.clear()
 
     logger.info(f"Deleted {count} matches.")
 
@@ -279,7 +352,21 @@ def delete_user_token_from_redis(
 
 if __name__ == "__main__":
     parser = argparse.ArgumentParser(description="Onyx Redis Manager")
-    parser.add_argument("--command", type=str, help="Operation to run", required=True)
+    parser.add_argument(
+        "--command",
+        type=OnyxRedisCommand,
+        help="The command to run",
+        choices=list(OnyxRedisCommand),
+        required=True,
+    )
+
+    parser.add_argument(
+        "--ssl",
+        type=bool,
+        default=REDIS_SSL,
+        help="Use SSL when connecting to Redis. Usually True for prod and False for local testing",
+        required=False,
+    )
 
     parser.add_argument(
         "--host",
@@ -342,6 +429,13 @@ if __name__ == "__main__":
         required=False,
     )
 
+    parser.add_argument(
+        "--cc-pair",
+        type=int,
+        help="A connector credential pair id. Used with the purge_deletion_locks command.",
+        required=False,
+    )
+
     args = parser.parse_args()
 
     if args.tenant_id:
@@ -368,10 +462,12 @@ if __name__ == "__main__":
         command=args.command,
         batch=args.batch,
         dry_run=args.dry_run,
+        ssl=args.ssl,
         host=args.host,
         port=args.port,
         db=args.db,
         password=args.password,
         user_email=args.user_email,
+        cc_pair_id=args.cc_pair,
     )
     sys.exit(exitcode)

backend/tests/integration/common_utils/constants.py

@@ -3,7 +3,7 @@ import os
 ADMIN_USER_NAME = "admin_user"
 
 API_SERVER_PROTOCOL = os.getenv("API_SERVER_PROTOCOL") or "http"
-API_SERVER_HOST = os.getenv("API_SERVER_HOST") or "localhost"
+API_SERVER_HOST = os.getenv("API_SERVER_HOST") or "127.0.0.1"
 API_SERVER_PORT = os.getenv("API_SERVER_PORT") or "8080"
 API_SERVER_URL = f"{API_SERVER_PROTOCOL}://{API_SERVER_HOST}:{API_SERVER_PORT}"
 MAX_DELAY = 45

backend/tests/integration/common_utils/managers/connector.py

@@ -30,8 +30,10 @@ class ConnectorManager:
             name=name,
             source=source,
             input_type=input_type,
-            connector_specific_config=connector_specific_config
-            or {"file_locations": []},
+            connector_specific_config=(
+                connector_specific_config
+                or ({"file_locations": []} if source == DocumentSource.FILE else {})
+            ),
             access_type=access_type,
             groups=groups or [],
         )

backend/tests/integration/common_utils/managers/user.py

@@ -88,8 +88,6 @@ class UserManager:
         if not session_cookie:
             raise Exception("Failed to login")
 
-        print(f"Logged in as {test_user.email}")
-
         # Set cookies in the headers
         test_user.headers["Cookie"] = f"fastapiusersauth={session_cookie}; "
         test_user.cookies = {"fastapiusersauth": session_cookie}

deployment/data/nginx/run-nginx.sh

@@ -1,5 +1,5 @@
 # fill in the template
-envsubst '$SSL_CERT_FILE_NAME $SSL_CERT_KEY_FILE_NAME' < "/etc/nginx/conf.d/$1" > /etc/nginx/conf.d/app.conf
+envsubst '$DOMAIN $SSL_CERT_FILE_NAME $SSL_CERT_KEY_FILE_NAME' < "/etc/nginx/conf.d/$1" > /etc/nginx/conf.d/app.conf
 
 # wait for the api_server to be ready
 echo "Waiting for API server to boot up; this may take a minute or two..."

deployment/docker_compose/docker-compose.dev.yml

@@ -36,6 +36,7 @@ services:
       - OPENID_CONFIG_URL=${OPENID_CONFIG_URL:-}
       - TRACK_EXTERNAL_IDP_EXPIRY=${TRACK_EXTERNAL_IDP_EXPIRY:-}
       - CORS_ALLOWED_ORIGIN=${CORS_ALLOWED_ORIGIN:-}
+      - INTEGRATION_TESTS_MODE=${INTEGRATION_TESTS_MODE:-}
       # Gen AI Settings
       - GEN_AI_MAX_TOKENS=${GEN_AI_MAX_TOKENS:-}
       - QA_TIMEOUT=${QA_TIMEOUT:-}

web/src/app/chat/ChatPage.tsx

@@ -47,6 +47,7 @@ import {
   removeMessage,
   sendMessage,
   setMessageAsLatest,
+  updateLlmOverrideForChatSession,
   updateParentChildren,
   uploadFilesForChat,
   useScrollonStream,
@@ -65,7 +66,7 @@ import {
 import { usePopup } from "@/components/admin/connectors/Popup";
 import { SEARCH_PARAM_NAMES, shouldSubmitOnLoad } from "./searchParams";
 import { useDocumentSelection } from "./useDocumentSelection";
-import { LlmOverride, useFilters, useLlmOverride } from "@/lib/hooks";
+import { LlmDescriptor, useFilters, useLlmManager } from "@/lib/hooks";
 import { ChatState, FeedbackType, RegenerationState } from "./types";
 import { DocumentResults } from "./documentSidebar/DocumentResults";
 import { OnyxInitializingLoader } from "@/components/OnyxInitializingLoader";
@@ -89,7 +90,11 @@ import {
 import { buildFilters } from "@/lib/search/utils";
 import { SettingsContext } from "@/components/settings/SettingsProvider";
 import Dropzone from "react-dropzone";
-import { checkLLMSupportsImageInput, getFinalLLM } from "@/lib/llm/utils";
+import {
+  checkLLMSupportsImageInput,
+  getFinalLLM,
+  structureValue,
+} from "@/lib/llm/utils";
 import { ChatInputBar } from "./input/ChatInputBar";
 import { useChatContext } from "@/components/context/ChatContext";
 import { v4 as uuidv4 } from "uuid";
@@ -194,16 +199,6 @@ export function ChatPage({
     return screenSize;
   }
 
-  const { height: screenHeight } = useScreenSize();
-
-  const getContainerHeight = () => {
-    if (autoScrollEnabled) return undefined;
-
-    if (screenHeight < 600) return "20vh";
-    if (screenHeight < 1200) return "30vh";
-    return "40vh";
-  };
-
   // handle redirect if chat page is disabled
   // NOTE: this must be done here, in a client component since
   // settings are passed in via Context and therefore aren't
@@ -222,6 +217,7 @@ export function ChatPage({
     setProSearchEnabled(!proSearchEnabled);
   };
 
+  const isInitialLoad = useRef(true);
   const [userSettingsToggled, setUserSettingsToggled] = useState(false);
 
   const {
@@ -356,7 +352,7 @@ export function ChatPage({
     ]
   );
 
-  const llmOverrideManager = useLlmOverride(
+  const llmManager = useLlmManager(
     llmProviders,
     selectedChatSession,
     liveAssistant
@@ -520,8 +516,17 @@ export function ChatPage({
       scrollInitialized.current = false;
 
       if (!hasPerformedInitialScroll) {
+        if (isInitialLoad.current) {
+          setHasPerformedInitialScroll(true);
+          isInitialLoad.current = false;
+        }
         clientScrollToBottom();
+
+        setTimeout(() => {
+          setHasPerformedInitialScroll(true);
+        }, 100);
       } else if (isChatSessionSwitch) {
+        setHasPerformedInitialScroll(true);
         clientScrollToBottom(true);
       }
 
@@ -1130,6 +1135,56 @@ export function ChatPage({
     });
   };
   const [uncaughtError, setUncaughtError] = useState<string | null>(null);
+  const [agenticGenerating, setAgenticGenerating] = useState(false);
+
+  const autoScrollEnabled =
+    (user?.preferences?.auto_scroll && !agenticGenerating) ?? false;
+
+  useScrollonStream({
+    chatState: currentSessionChatState,
+    scrollableDivRef,
+    scrollDist,
+    endDivRef,
+    debounceNumber,
+    mobile: settings?.isMobile,
+    enableAutoScroll: autoScrollEnabled,
+  });
+
+  // Track whether a message has been sent during this page load, keyed by chat session id
+  const [sessionHasSentLocalUserMessage, setSessionHasSentLocalUserMessage] =
+    useState<Map<string | null, boolean>>(new Map());
+
+  // Update the local state for a session once the user sends a message
+  const markSessionMessageSent = (sessionId: string | null) => {
+    setSessionHasSentLocalUserMessage((prev) => {
+      const newMap = new Map(prev);
+      newMap.set(sessionId, true);
+      return newMap;
+    });
+  };
+  const currentSessionHasSentLocalUserMessage = useMemo(
+    () => (sessionId: string | null) => {
+      return sessionHasSentLocalUserMessage.size === 0
+        ? undefined
+        : sessionHasSentLocalUserMessage.get(sessionId) || false;
+    },
+    [sessionHasSentLocalUserMessage]
+  );
+
+  const { height: screenHeight } = useScreenSize();
+
+  const getContainerHeight = useMemo(() => {
+    return () => {
+      if (!currentSessionHasSentLocalUserMessage(chatSessionIdRef.current)) {
+        return undefined;
+      }
+      if (autoScrollEnabled) return undefined;
+
+      if (screenHeight < 600) return "40vh";
+      if (screenHeight < 1200) return "50vh";
+      return "60vh";
+    };
+  }, [autoScrollEnabled, screenHeight, currentSessionHasSentLocalUserMessage]);
 
   const onSubmit = async ({
     messageIdToResend,
@@ -1138,7 +1193,7 @@ export function ChatPage({
     forceSearch,
     isSeededChat,
     alternativeAssistantOverride = null,
-    modelOverRide,
+    modelOverride,
     regenerationRequest,
     overrideFileDescriptors,
   }: {
@@ -1148,7 +1203,7 @@ export function ChatPage({
     forceSearch?: boolean;
     isSeededChat?: boolean;
     alternativeAssistantOverride?: Persona | null;
-    modelOverRide?: LlmOverride;
+    modelOverride?: LlmDescriptor;
     regenerationRequest?: RegenerationRequest | null;
     overrideFileDescriptors?: FileDescriptor[];
   } = {}) => {
@@ -1156,6 +1211,9 @@ export function ChatPage({
     let frozenSessionId = currentSessionId();
     updateCanContinue(false, frozenSessionId);
 
+    // Mark that we've sent a message for this session in the current page load
+    markSessionMessageSent(frozenSessionId);
+
     if (currentChatState() != "input") {
       if (currentChatState() == "uploading") {
         setPopup({
@@ -1191,6 +1249,22 @@ export function ChatPage({
       currChatSessionId = chatSessionIdRef.current as string;
     }
     frozenSessionId = currChatSessionId;
+    // update the selected model for the chat session if one is specified so that
+    // it persists across page reloads. Do not `await` here so that the message
+    // request can continue and this will just happen in the background.
+    // NOTE: only set the model override for the chat session once we send a
+    // message with it. If the user switches models and then starts a new
+    // chat session, it is unexpected for that model to be used when they
+    // return to this session the next day.
+    let finalLLM = modelOverride || llmManager.currentLlm;
+    updateLlmOverrideForChatSession(
+      currChatSessionId,
+      structureValue(
+        finalLLM.name || "",
+        finalLLM.provider || "",
+        finalLLM.modelName || ""
+      )
+    );
 
     updateStatesWithNewSessionId(currChatSessionId);
 
@@ -1250,11 +1324,14 @@ export function ChatPage({
         : null) ||
       (messageMap.size === 1 ? Array.from(messageMap.values())[0] : null);
 
-    const currentAssistantId = alternativeAssistantOverride
-      ? alternativeAssistantOverride.id
-      : alternativeAssistant
-        ? alternativeAssistant.id
-        : liveAssistant.id;
+    let currentAssistantId;
+    if (alternativeAssistantOverride) {
+      currentAssistantId = alternativeAssistantOverride.id;
+    } else if (alternativeAssistant) {
+      currentAssistantId = alternativeAssistant.id;
+    } else {
+      currentAssistantId = liveAssistant.id;
+    }
 
     resetInputBar();
     let messageUpdates: Message[] | null = null;
@@ -1326,15 +1403,13 @@ export function ChatPage({
         forceSearch,
         regenerate: regenerationRequest !== undefined,
         modelProvider:
-          modelOverRide?.name ||
-          llmOverrideManager.llmOverride.name ||
-          undefined,
+          modelOverride?.name || llmManager.currentLlm.name || undefined,
         modelVersion:
-          modelOverRide?.modelName ||
-          llmOverrideManager.llmOverride.modelName ||
+          modelOverride?.modelName ||
+          llmManager.currentLlm.modelName ||
           searchParams.get(SEARCH_PARAM_NAMES.MODEL_VERSION) ||
           undefined,
-        temperature: llmOverrideManager.temperature || undefined,
+        temperature: llmManager.temperature || undefined,
         systemPromptOverride:
           searchParams.get(SEARCH_PARAM_NAMES.SYSTEM_PROMPT) || undefined,
         useExistingUserMessage: isSeededChat,
@@ -1802,7 +1877,7 @@ export function ChatPage({
     const [_, llmModel] = getFinalLLM(
       llmProviders,
       liveAssistant,
-      llmOverrideManager.llmOverride
+      llmManager.currentLlm
     );
     const llmAcceptsImages = checkLLMSupportsImageInput(llmModel);
 
@@ -1857,7 +1932,6 @@ export function ChatPage({
   // Used to maintain a "time out" for history sidebar so our existing refs can have time to process change
   const [untoggled, setUntoggled] = useState(false);
   const [loadingError, setLoadingError] = useState<string | null>(null);
-  const [agenticGenerating, setAgenticGenerating] = useState(false);
 
   const explicitlyUntoggle = () => {
     setShowHistorySidebar(false);
@@ -1899,19 +1973,6 @@ export function ChatPage({
     isAnonymousUser: user?.is_anonymous_user,
   });
 
-  const autoScrollEnabled =
-    (user?.preferences?.auto_scroll && !agenticGenerating) ?? false;
-
-  useScrollonStream({
-    chatState: currentSessionChatState,
-    scrollableDivRef,
-    scrollDist,
-    endDivRef,
-    debounceNumber,
-    mobile: settings?.isMobile,
-    enableAutoScroll: autoScrollEnabled,
-  });
-
   // Virtualization + Scrolling related effects and functions
   const scrollInitialized = useRef(false);
   interface VisibleRange {
@@ -2121,7 +2182,7 @@ export function ChatPage({
   }, [searchParams, router]);
 
   useEffect(() => {
-    llmOverrideManager.updateImageFilesPresent(imageFileInMessageHistory);
+    llmManager.updateImageFilesPresent(imageFileInMessageHistory);
   }, [imageFileInMessageHistory]);
 
   const pathname = usePathname();
@@ -2175,9 +2236,9 @@ export function ChatPage({
 
   function createRegenerator(regenerationRequest: RegenerationRequest) {
     // Returns new function that only needs `modelOverRide` to be specified when called
-    return async function (modelOverRide: LlmOverride) {
+    return async function (modelOverride: LlmDescriptor) {
       return await onSubmit({
-        modelOverRide,
+        modelOverride,
         messageIdToResend: regenerationRequest.parentMessage.messageId,
         regenerationRequest,
         forceSearch: regenerationRequest.forceSearch,
@@ -2258,9 +2319,7 @@ export function ChatPage({
       {(settingsToggled || userSettingsToggled) && (
         <UserSettingsModal
           setPopup={setPopup}
-          setLlmOverride={(newOverride) =>
-            llmOverrideManager.updateLLMOverride(newOverride)
-          }
+          setCurrentLlm={(newLlm) => llmManager.updateCurrentLlm(newLlm)}
           defaultModel={user?.preferences.default_model!}
           llmProviders={llmProviders}
           onClose={() => {
@@ -2324,7 +2383,7 @@ export function ChatPage({
         <ShareChatSessionModal
           assistantId={liveAssistant?.id}
           message={message}
-          modelOverride={llmOverrideManager.llmOverride}
+          modelOverride={llmManager.currentLlm}
           chatSessionId={sharedChatSession.id}
           existingSharedStatus={sharedChatSession.shared_status}
           onClose={() => setSharedChatSession(null)}
@@ -2342,7 +2401,7 @@ export function ChatPage({
         <ShareChatSessionModal
           message={message}
           assistantId={liveAssistant?.id}
-          modelOverride={llmOverrideManager.llmOverride}
+          modelOverride={llmManager.currentLlm}
           chatSessionId={chatSessionIdRef.current}
           existingSharedStatus={chatSessionSharedStatus}
           onClose={() => setSharingModalVisible(false)}
@@ -2572,6 +2631,7 @@ export function ChatPage({
                             style={{ overflowAnchor: "none" }}
                             key={currentSessionId()}
                             className={
+                              (hasPerformedInitialScroll ? "" : " hidden ") +
                               "desktop:-ml-4 w-full mx-auto " +
                               "absolute mobile:top-0 desktop:top-0 left-0 " +
                               (settings?.enterpriseSettings
@@ -3058,7 +3118,7 @@ export function ChatPage({
                                               messageId: message.messageId,
                                               parentMessage: parentMessage!,
                                               forceSearch: true,
-                                            })(llmOverrideManager.llmOverride);
+                                            })(llmManager.currentLlm);
                                           } else {
                                             setPopup({
                                               type: "error",
@@ -3203,7 +3263,7 @@ export function ChatPage({
                               availableDocumentSets={documentSets}
                               availableTags={tags}
                               filterManager={filterManager}
-                              llmOverrideManager={llmOverrideManager}
+                              llmManager={llmManager}
                               removeDocs={() => {
                                 clearSelectedDocuments();
                               }}

web/src/app/chat/RegenerateOption.tsx

@@ -1,8 +1,8 @@
 import { useChatContext } from "@/components/context/ChatContext";
 import {
   getDisplayNameForModel,
-  LlmOverride,
-  useLlmOverride,
+  LlmDescriptor,
+  useLlmManager,
 } from "@/lib/hooks";
 import { StringOrNumberOption } from "@/components/Dropdown";
 
@@ -106,13 +106,13 @@ export default function RegenerateOption({
   onDropdownVisibleChange,
 }: {
   selectedAssistant: Persona;
-  regenerate: (modelOverRide: LlmOverride) => Promise<void>;
+  regenerate: (modelOverRide: LlmDescriptor) => Promise<void>;
   overriddenModel?: string;
   onHoverChange: (isHovered: boolean) => void;
   onDropdownVisibleChange: (isVisible: boolean) => void;
 }) {
   const { llmProviders } = useChatContext();
-  const llmOverrideManager = useLlmOverride(llmProviders);
+  const llmManager = useLlmManager(llmProviders);
 
   const [_, llmName] = getFinalLLM(llmProviders, selectedAssistant, null);
 
@@ -148,7 +148,7 @@ export default function RegenerateOption({
   );
 
   const currentModelName =
-    llmOverrideManager?.llmOverride.modelName ||
+    llmManager?.currentLlm.modelName ||
     (selectedAssistant
       ? selectedAssistant.llm_model_version_override || llmName
       : llmName);

web/src/app/chat/input/ChatInputBar.tsx

@@ -6,7 +6,7 @@ import { Persona } from "@/app/admin/assistants/interfaces";
 import LLMPopover from "./LLMPopover";
 import { InputPrompt } from "@/app/chat/interfaces";
 
-import { FilterManager, LlmOverrideManager } from "@/lib/hooks";
+import { FilterManager, LlmManager } from "@/lib/hooks";
 import { useChatContext } from "@/components/context/ChatContext";
 import { ChatFileType, FileDescriptor } from "../interfaces";
 import {
@@ -180,7 +180,7 @@ interface ChatInputBarProps {
   setMessage: (message: string) => void;
   stopGenerating: () => void;
   onSubmit: () => void;
-  llmOverrideManager: LlmOverrideManager;
+  llmManager: LlmManager;
   chatState: ChatState;
   alternativeAssistant: Persona | null;
   // assistants
@@ -225,7 +225,7 @@ export function ChatInputBar({
   availableSources,
   availableDocumentSets,
   availableTags,
-  llmOverrideManager,
+  llmManager,
   proSearchEnabled,
   setProSearchEnabled,
 }: ChatInputBarProps) {
@@ -781,7 +781,7 @@ export function ChatInputBar({
 
                 <LLMPopover
                   llmProviders={llmProviders}
-                  llmOverrideManager={llmOverrideManager}
+                  llmManager={llmManager}
                   requiresImageGeneration={false}
                   currentAssistant={selectedAssistant}
                 />

web/src/app/chat/input/LLMPopover.tsx

@@ -16,7 +16,7 @@ import {
   LLMProviderDescriptor,
 } from "@/app/admin/configuration/llm/interfaces";
 import { Persona } from "@/app/admin/assistants/interfaces";
-import { LlmOverrideManager } from "@/lib/hooks";
+import { LlmManager } from "@/lib/hooks";
 
 import {
   Tooltip,
@@ -31,21 +31,19 @@ import { useUser } from "@/components/user/UserProvider";
 
 interface LLMPopoverProps {
   llmProviders: LLMProviderDescriptor[];
-  llmOverrideManager: LlmOverrideManager;
+  llmManager: LlmManager;
   requiresImageGeneration?: boolean;
   currentAssistant?: Persona;
 }
 
 export default function LLMPopover({
   llmProviders,
-  llmOverrideManager,
+  llmManager,
   requiresImageGeneration,
   currentAssistant,
 }: LLMPopoverProps) {
   const [isOpen, setIsOpen] = useState(false);
   const { user } = useUser();
-  const { llmOverride, updateLLMOverride } = llmOverrideManager;
-  const currentLlm = llmOverride.modelName;
 
   const llmOptionsByProvider: {
     [provider: string]: {
@@ -93,19 +91,19 @@ export default function LLMPopover({
     : null;
 
   const [localTemperature, setLocalTemperature] = useState(
-    llmOverrideManager.temperature ?? 0.5
+    llmManager.temperature ?? 0.5
   );
 
   useEffect(() => {
-    setLocalTemperature(llmOverrideManager.temperature ?? 0.5);
-  }, [llmOverrideManager.temperature]);
+    setLocalTemperature(llmManager.temperature ?? 0.5);
+  }, [llmManager.temperature]);
 
   const handleTemperatureChange = (value: number[]) => {
     setLocalTemperature(value[0]);
   };
 
   const handleTemperatureChangeComplete = (value: number[]) => {
-    llmOverrideManager.updateTemperature(value[0]);
+    llmManager.updateTemperature(value[0]);
   };
 
   return (
@@ -120,15 +118,15 @@ export default function LLMPopover({
             toggle
             flexPriority="stiff"
             name={getDisplayNameForModel(
-              llmOverrideManager?.llmOverride.modelName ||
+              llmManager?.currentLlm.modelName ||
                 defaultModelDisplayName ||
                 "Models"
             )}
             Icon={getProviderIcon(
-              llmOverrideManager?.llmOverride.provider ||
+              llmManager?.currentLlm.provider ||
                 defaultProvider?.provider ||
                 "anthropic",
-              llmOverrideManager?.llmOverride.modelName ||
+              llmManager?.currentLlm.modelName ||
                 defaultProvider?.default_model_name ||
                 "claude-3-5-sonnet-20240620"
             )}
@@ -147,12 +145,12 @@ export default function LLMPopover({
                 <button
                   key={index}
                   className={`w-full flex items-center gap-x-2 px-3 py-2 text-sm text-left hover:bg-background-100 dark:hover:bg-neutral-800 transition-colors duration-150 ${
-                    currentLlm === name
+                    llmManager.currentLlm.modelName === name
                       ? "bg-background-100 dark:bg-neutral-900 text-text"
                       : "text-text-darker"
                   }`}
                   onClick={() => {
-                    updateLLMOverride(destructureValue(value));
+                    llmManager.updateCurrentLlm(destructureValue(value));
                     setIsOpen(false);
                   }}
                 >
@@ -172,7 +170,7 @@ export default function LLMPopover({
                       );
                     }
                   })()}
-                  {llmOverrideManager.imageFilesPresent &&
+                  {llmManager.imageFilesPresent &&
                     !checkLLMSupportsImageInput(name) && (
                       <TooltipProvider>
                         <Tooltip delayDuration={0}>
@@ -199,7 +197,7 @@ export default function LLMPopover({
             <div className="w-full px-3 py-2">
               <Slider
                 value={[localTemperature]}
-                max={llmOverrideManager.maxTemperature}
+                max={llmManager.maxTemperature}
                 min={0}
                 step={0.01}
                 onValueChange={handleTemperatureChange}

web/src/app/chat/lib.tsx

@@ -65,7 +65,7 @@ export function getChatRetentionInfo(
   };
 }
 
-export async function updateModelOverrideForChatSession(
+export async function updateLlmOverrideForChatSession(
   chatSessionId: string,
   newAlternateModel: string
 ) {
@@ -236,7 +236,7 @@ export async function* sendMessage({
           }
         : null,
     use_existing_user_message: useExistingUserMessage,
-    use_agentic_search: useLanggraph,
+    use_agentic_search: useLanggraph ?? false,
   });
 
   const response = await fetch(`/api/chat/send-message`, {

web/src/app/chat/message/AgenticMessage.tsx

@@ -44,7 +44,7 @@ import { ValidSources } from "@/lib/types";
 import { useMouseTracking } from "./hooks";
 import { SettingsContext } from "@/components/settings/SettingsProvider";
 import RegenerateOption from "../RegenerateOption";
-import { LlmOverride } from "@/lib/hooks";
+import { LlmDescriptor } from "@/lib/hooks";
 import { ContinueGenerating } from "./ContinueMessage";
 import { MemoizedAnchor, MemoizedParagraph } from "./MemoizedTextComponents";
 import { extractCodeText, preprocessLaTeX } from "./codeUtils";
@@ -117,7 +117,7 @@ export const AgenticMessage = ({
   isComplete?: boolean;
   handleFeedback?: (feedbackType: FeedbackType) => void;
   overriddenModel?: string;
-  regenerate?: (modelOverRide: LlmOverride) => Promise<void>;
+  regenerate?: (modelOverRide: LlmDescriptor) => Promise<void>;
   setPresentingDocument?: (document: OnyxDocument) => void;
   toggleDocDisplay?: (agentic: boolean) => void;
   error?: string | null;

web/src/app/chat/message/Messages.tsx

@@ -58,7 +58,7 @@ import { useMouseTracking } from "./hooks";
 import { SettingsContext } from "@/components/settings/SettingsProvider";
 import GeneratingImageDisplay from "../tools/GeneratingImageDisplay";
 import RegenerateOption from "../RegenerateOption";
-import { LlmOverride } from "@/lib/hooks";
+import { LlmDescriptor } from "@/lib/hooks";
 import { ContinueGenerating } from "./ContinueMessage";
 import { MemoizedAnchor, MemoizedParagraph } from "./MemoizedTextComponents";
 import { extractCodeText, preprocessLaTeX } from "./codeUtils";
@@ -213,7 +213,7 @@ export const AIMessage = ({
   handleForceSearch?: () => void;
   retrievalDisabled?: boolean;
   overriddenModel?: string;
-  regenerate?: (modelOverRide: LlmOverride) => Promise<void>;
+  regenerate?: (modelOverRide: LlmDescriptor) => Promise<void>;
   setPresentingDocument: (document: OnyxDocument) => void;
   removePadding?: boolean;
 }) => {

web/src/app/chat/modal/ShareChatSessionModal.tsx

@@ -11,7 +11,7 @@ import { CopyButton } from "@/components/CopyButton";
 import { SEARCH_PARAM_NAMES } from "../searchParams";
 import { usePopup } from "@/components/admin/connectors/Popup";
 import { structureValue } from "@/lib/llm/utils";
-import { LlmOverride } from "@/lib/hooks";
+import { LlmDescriptor } from "@/lib/hooks";
 import { Separator } from "@/components/ui/separator";
 import { AdvancedOptionsToggle } from "@/components/AdvancedOptionsToggle";
 
@@ -38,7 +38,7 @@ async function generateShareLink(chatSessionId: string) {
 async function generateSeedLink(
   message?: string,
   assistantId?: number,
-  modelOverride?: LlmOverride
+  modelOverride?: LlmDescriptor
 ) {
   const baseUrl = `${window.location.protocol}//${window.location.host}`;
   const model = modelOverride
@@ -92,7 +92,7 @@ export function ShareChatSessionModal({
   onClose: () => void;
   message?: string;
   assistantId?: number;
-  modelOverride?: LlmOverride;
+  modelOverride?: LlmDescriptor;
 }) {
   const [shareLink, setShareLink] = useState<string>(
     existingSharedStatus === ChatSessionSharedStatus.Public

web/src/app/chat/modal/UserSettingsModal.tsx

@@ -1,6 +1,6 @@
 import { useContext, useEffect, useRef, useState } from "react";
 import { Modal } from "@/components/Modal";
-import { getDisplayNameForModel, LlmOverride } from "@/lib/hooks";
+import { getDisplayNameForModel, LlmDescriptor } from "@/lib/hooks";
 import { LLMProviderDescriptor } from "@/app/admin/configuration/llm/interfaces";
 
 import { destructureValue, structureValue } from "@/lib/llm/utils";
@@ -31,12 +31,12 @@ export function UserSettingsModal({
   setPopup,
   llmProviders,
   onClose,
-  setLlmOverride,
+  setCurrentLlm,
   defaultModel,
 }: {
   setPopup: (popupSpec: PopupSpec | null) => void;
   llmProviders: LLMProviderDescriptor[];
-  setLlmOverride?: (newOverride: LlmOverride) => void;
+  setCurrentLlm?: (newLlm: LlmDescriptor) => void;
   onClose: () => void;
   defaultModel: string | null;
 }) {
@@ -127,18 +127,14 @@ export function UserSettingsModal({
     );
   });
 
-  const llmOptions = Object.entries(llmOptionsByProvider).flatMap(
-    ([provider, options]) => [...options]
-  );
-
   const router = useRouter();
   const handleChangedefaultModel = async (defaultModel: string | null) => {
     try {
       const response = await setUserDefaultModel(defaultModel);
 
       if (response.ok) {
-        if (defaultModel && setLlmOverride) {
-          setLlmOverride(destructureValue(defaultModel));
+        if (defaultModel && setCurrentLlm) {
+          setCurrentLlm(destructureValue(defaultModel));
         }
         setPopup({
           message: "Default model updated successfully",

web/src/lib/hooks.ts

@@ -360,18 +360,18 @@ export const useUsers = ({ includeApiKeys }: UseUsersParams) => {
   };
 };
 
-export interface LlmOverride {
+export interface LlmDescriptor {
   name: string;
   provider: string;
   modelName: string;
 }
 
-export interface LlmOverrideManager {
-  llmOverride: LlmOverride;
-  updateLLMOverride: (newOverride: LlmOverride) => void;
+export interface LlmManager {
+  currentLlm: LlmDescriptor;
+  updateCurrentLlm: (newOverride: LlmDescriptor) => void;
   temperature: number;
   updateTemperature: (temperature: number) => void;
-  updateModelOverrideForChatSession: (chatSession?: ChatSession) => void;
+  updateModelOverrideBasedOnChatSession: (chatSession?: ChatSession) => void;
   imageFilesPresent: boolean;
   updateImageFilesPresent: (present: boolean) => void;
   liveAssistant: Persona | null;
@@ -400,7 +400,7 @@ Thus, the input should be
 
 Changes take place as
 - liveAssistant or currentChatSession changes (and the associated model override is set)
-- (uploadLLMOverride) User explicitly setting a model override (and we explicitly override and set the userSpecifiedOverride which we'll use in place of the user preferences unless overridden by an assistant)
+- (updateCurrentLlm) User explicitly setting a model override (and we explicitly override and set the userSpecifiedOverride which we'll use in place of the user preferences unless overridden by an assistant)
 
 If we have a live assistant, we should use that model override
 
@@ -419,55 +419,78 @@ This approach ensures that user preferences are maintained for existing chats wh
 providing appropriate defaults for new conversations based on the available tools.
 */
 
-export function useLlmOverride(
+export function useLlmManager(
   llmProviders: LLMProviderDescriptor[],
   currentChatSession?: ChatSession,
   liveAssistant?: Persona
-): LlmOverrideManager {
+): LlmManager {
   const { user } = useUser();
 
+  const [userHasManuallyOverriddenLLM, setUserHasManuallyOverriddenLLM] =
+    useState(false);
   const [chatSession, setChatSession] = useState<ChatSession | null>(null);
+  const [currentLlm, setCurrentLlm] = useState<LlmDescriptor>({
+    name: "",
+    provider: "",
+    modelName: "",
+  });
 
-  const llmOverrideUpdate = () => {
-    if (liveAssistant?.llm_model_version_override) {
-      setLlmOverride(
-        getValidLlmOverride(liveAssistant.llm_model_version_override)
-      );
-    } else if (currentChatSession?.current_alternate_model) {
-      setLlmOverride(
-        getValidLlmOverride(currentChatSession.current_alternate_model)
-      );
-    } else if (user?.preferences?.default_model) {
-      setLlmOverride(getValidLlmOverride(user.preferences.default_model));
-      return;
-    } else {
-      const defaultProvider = llmProviders.find(
-        (provider) => provider.is_default_provider
-      );
+  const llmUpdate = () => {
+    /* Should be called when the live assistant or current chat session changes */
 
-      if (defaultProvider) {
-        setLlmOverride({
-          name: defaultProvider.name,
-          provider: defaultProvider.provider,
-          modelName: defaultProvider.default_model_name,
-        });
+    // separate function so we can `return` to break out
+    const _llmUpdate = () => {
+      // if the user has overridden in this session and just switched to a brand
+      // new session, use their manually specified model
+      if (userHasManuallyOverriddenLLM && !currentChatSession) {
+        return;
       }
-    }
+
+      if (currentChatSession?.current_alternate_model) {
+        setCurrentLlm(
+          getValidLlmDescriptor(currentChatSession.current_alternate_model)
+        );
+      } else if (liveAssistant?.llm_model_version_override) {
+        setCurrentLlm(
+          getValidLlmDescriptor(liveAssistant.llm_model_version_override)
+        );
+      } else if (userHasManuallyOverriddenLLM) {
+        // if the user has an override and there's nothing special about the
+        // current chat session, use the override
+        return;
+      } else if (user?.preferences?.default_model) {
+        setCurrentLlm(getValidLlmDescriptor(user.preferences.default_model));
+      } else {
+        const defaultProvider = llmProviders.find(
+          (provider) => provider.is_default_provider
+        );
+
+        if (defaultProvider) {
+          setCurrentLlm({
+            name: defaultProvider.name,
+            provider: defaultProvider.provider,
+            modelName: defaultProvider.default_model_name,
+          });
+        }
+      }
+    };
+
+    _llmUpdate();
     setChatSession(currentChatSession || null);
   };
 
-  const getValidLlmOverride = (
-    overrideModel: string | null | undefined
-  ): LlmOverride => {
-    if (overrideModel) {
-      const model = destructureValue(overrideModel);
+  const getValidLlmDescriptor = (
+    modelName: string | null | undefined
+  ): LlmDescriptor => {
+    if (modelName) {
+      const model = destructureValue(modelName);
       if (!(model.modelName && model.modelName.length > 0)) {
         const provider = llmProviders.find((p) =>
-          p.model_names.includes(overrideModel)
+          p.model_names.includes(modelName)
         );
         if (provider) {
           return {
-            modelName: overrideModel,
+            modelName: modelName,
             name: provider.name,
             provider: provider.provider,
           };
@@ -491,38 +514,32 @@ export function useLlmOverride(
     setImageFilesPresent(present);
   };
 
-  const [llmOverride, setLlmOverride] = useState<LlmOverride>({
-    name: "",
-    provider: "",
-    modelName: "",
-  });
-
-  // Manually set the override
-  const updateLLMOverride = (newOverride: LlmOverride) => {
+  // Manually set the LLM
+  const updateCurrentLlm = (newLlm: LlmDescriptor) => {
     const provider =
-      newOverride.provider ||
-      findProviderForModel(llmProviders, newOverride.modelName);
+      newLlm.provider || findProviderForModel(llmProviders, newLlm.modelName);
     const structuredValue = structureValue(
-      newOverride.name,
+      newLlm.name,
       provider,
-      newOverride.modelName
+      newLlm.modelName
     );
-    setLlmOverride(getValidLlmOverride(structuredValue));
+    setCurrentLlm(getValidLlmDescriptor(structuredValue));
+    setUserHasManuallyOverriddenLLM(true);
   };
 
-  const updateModelOverrideForChatSession = (chatSession?: ChatSession) => {
+  const updateModelOverrideBasedOnChatSession = (chatSession?: ChatSession) => {
     if (chatSession && chatSession.current_alternate_model?.length > 0) {
-      setLlmOverride(getValidLlmOverride(chatSession.current_alternate_model));
+      setCurrentLlm(getValidLlmDescriptor(chatSession.current_alternate_model));
     }
   };
 
   const [temperature, setTemperature] = useState<number>(() => {
-    llmOverrideUpdate();
+    llmUpdate();
 
     if (currentChatSession?.current_temperature_override != null) {
       return Math.min(
         currentChatSession.current_temperature_override,
-        isAnthropic(llmOverride.provider, llmOverride.modelName) ? 1.0 : 2.0
+        isAnthropic(currentLlm.provider, currentLlm.modelName) ? 1.0 : 2.0
       );
     } else if (
       liveAssistant?.tools.some((tool) => tool.name === SEARCH_TOOL_ID)
@@ -533,22 +550,23 @@ export function useLlmOverride(
   });
 
   const maxTemperature = useMemo(() => {
-    return isAnthropic(llmOverride.provider, llmOverride.modelName) ? 1.0 : 2.0;
-  }, [llmOverride]);
+    return isAnthropic(currentLlm.provider, currentLlm.modelName) ? 1.0 : 2.0;
+  }, [currentLlm]);
 
   useEffect(() => {
-    if (isAnthropic(llmOverride.provider, llmOverride.modelName)) {
+    if (isAnthropic(currentLlm.provider, currentLlm.modelName)) {
       const newTemperature = Math.min(temperature, 1.0);
       setTemperature(newTemperature);
       if (chatSession?.id) {
         updateTemperatureOverrideForChatSession(chatSession.id, newTemperature);
       }
     }
-  }, [llmOverride]);
+  }, [currentLlm]);
 
   useEffect(() => {
+    llmUpdate();
+
     if (!chatSession && currentChatSession) {
-      setChatSession(currentChatSession || null);
       if (temperature) {
         updateTemperatureOverrideForChatSession(
           currentChatSession.id,
@@ -570,7 +588,7 @@ export function useLlmOverride(
   }, [liveAssistant, currentChatSession]);
 
   const updateTemperature = (temperature: number) => {
-    if (isAnthropic(llmOverride.provider, llmOverride.modelName)) {
+    if (isAnthropic(currentLlm.provider, currentLlm.modelName)) {
       setTemperature((prevTemp) => Math.min(temperature, 1.0));
     } else {
       setTemperature(temperature);
@@ -581,9 +599,9 @@ export function useLlmOverride(
   };
 
   return {
-    updateModelOverrideForChatSession,
-    llmOverride,
-    updateLLMOverride,
+    updateModelOverrideBasedOnChatSession,
+    currentLlm,
+    updateCurrentLlm,
     temperature,
     updateTemperature,
     imageFilesPresent,

web/src/lib/llm/utils.ts

@@ -1,11 +1,11 @@
 import { Persona } from "@/app/admin/assistants/interfaces";
 import { LLMProviderDescriptor } from "@/app/admin/configuration/llm/interfaces";
-import { LlmOverride } from "@/lib/hooks";
+import { LlmDescriptor } from "@/lib/hooks";
 
 export function getFinalLLM(
   llmProviders: LLMProviderDescriptor[],
   persona: Persona | null,
-  llmOverride: LlmOverride | null
+  currentLlm: LlmDescriptor | null
 ): [string, string] {
   const defaultProvider = llmProviders.find(
     (llmProvider) => llmProvider.is_default_provider
@@ -26,9 +26,9 @@ export function getFinalLLM(
     model = persona.llm_model_version_override || model;
   }
 
-  if (llmOverride) {
-    provider = llmOverride.provider || provider;
-    model = llmOverride.modelName || model;
+  if (currentLlm) {
+    provider = currentLlm.provider || provider;
+    model = currentLlm.modelName || model;
   }
 
   return [provider, model];
@@ -37,7 +37,7 @@ export function getFinalLLM(
 export function getLLMProviderOverrideForPersona(
   liveAssistant: Persona,
   llmProviders: LLMProviderDescriptor[]
-): LlmOverride | null {
+): LlmDescriptor | null {
   const overrideProvider = liveAssistant.llm_model_provider_override;
   const overrideModel = liveAssistant.llm_model_version_override;
 
@@ -135,7 +135,7 @@ export const structureValue = (
   return `${name}__${provider}__${modelName}`;
 };
 
-export const destructureValue = (value: string): LlmOverride => {
+export const destructureValue = (value: string): LlmDescriptor => {
   const [displayName, provider, modelName] = value.split("__");
   return {
     name: displayName,

web/src/lib/users/UserSettings.tsx

@@ -1,5 +1,3 @@
-import { LlmOverride } from "../hooks";
-
 export async function setUserDefaultModel(
   model: string | null
 ): Promise<Response> {