k

* temporarily disabling validate indexing fences

* add back a few startup checks in the cloud

* use common vespa client to perform health check

* log vespa url and try using http1 on light worker index methods

---------

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

backend/ee/onyx/utils/telemetry.py

@@ -1,3 +1,5 @@
+from typing import Any
+
 from posthog import Posthog
 
 from ee.onyx.configs.app_configs import POSTHOG_API_KEY
@@ -6,13 +8,31 @@ from onyx.utils.logger import setup_logger
 
 logger = setup_logger()
 
-posthog = Posthog(project_api_key=POSTHOG_API_KEY, host=POSTHOG_HOST)
+
+def posthog_on_error(error: Any, items: Any) -> None:
+    logger.error(f"PostHog error: {error}, items: {items}")
+
+
+posthog = Posthog(
+    project_api_key=POSTHOG_API_KEY,
+    host=POSTHOG_HOST,
+    debug=True,
+    on_error=posthog_on_error,
+)
 
 
 def event_telemetry(
-    distinct_id: str,
-    event: str,
-    properties: dict | None = None,
+    distinct_id: str, event: str, properties: dict | None = None
 ) -> None:
     logger.info(f"Capturing Posthog event: {distinct_id} {event} {properties}")
-    posthog.capture(distinct_id, event, properties)
+    print("API KEY", POSTHOG_API_KEY)
+    print("HOST", POSTHOG_HOST)
+    try:
+        print(type(distinct_id))
+        print(type(event))
+        print(type(properties))
+        response = posthog.capture(distinct_id, event, properties)
+        posthog.flush()
+        print(response)
+    except Exception as e:
+        logger.error(f"Error capturing Posthog event: {e}")

backend/onyx/background/celery/apps/app_base.py

@@ -3,7 +3,6 @@ import multiprocessing
 import time
 from typing import Any
 
-import requests
 import sentry_sdk
 from celery import Task
 from celery.app import trace
@@ -23,6 +22,7 @@ from onyx.background.celery.apps.task_formatters import CeleryTaskPlainFormatter
 from onyx.background.celery.celery_utils import celery_is_worker_primary
 from onyx.configs.constants import OnyxRedisLocks
 from onyx.db.engine import get_sqlalchemy_engine
+from onyx.document_index.vespa.shared_utils.utils import get_vespa_http_client
 from onyx.document_index.vespa_constants import VESPA_CONFIG_SERVER_URL
 from onyx.redis.redis_connector import RedisConnector
 from onyx.redis.redis_connector_credential_pair import RedisConnectorCredentialPair
@@ -262,7 +262,8 @@ def wait_for_vespa(sender: Any, **kwargs: Any) -> None:
     logger.info("Vespa: Readiness probe starting.")
     while True:
         try:
-            response = requests.get(f"{VESPA_CONFIG_SERVER_URL}/state/v1/health")
+            client = get_vespa_http_client()
+            response = client.get(f"{VESPA_CONFIG_SERVER_URL}/state/v1/health")
             response.raise_for_status()
 
             response_dict = response.json()

backend/onyx/background/celery/apps/beat.py

@@ -13,7 +13,6 @@ from onyx.db.engine import SqlEngine
 from onyx.utils.logger import setup_logger
 from onyx.utils.variable_functionality import fetch_versioned_implementation
 from shared_configs.configs import IGNORED_SYNCING_TENANT_LIST
-from shared_configs.configs import MULTI_TENANT
 
 logger = setup_logger(__name__)
 
@@ -154,10 +153,6 @@ def on_beat_init(sender: Any, **kwargs: Any) -> None:
     SqlEngine.set_app_name(POSTGRES_CELERY_BEAT_APP_NAME)
     SqlEngine.init_engine(pool_size=2, max_overflow=0)
 
-    # Startup checks are not needed in multi-tenant case
-    if MULTI_TENANT:
-        return
-
     app_base.wait_for_redis(sender, **kwargs)
 
 

backend/onyx/background/celery/apps/heavy.py

@@ -61,13 +61,14 @@ def on_worker_init(sender: Any, **kwargs: Any) -> None:
     SqlEngine.set_app_name(POSTGRES_CELERY_WORKER_HEAVY_APP_NAME)
     SqlEngine.init_engine(pool_size=4, max_overflow=12)
 
-    # Startup checks are not needed in multi-tenant case
-    if MULTI_TENANT:
-        return
-
     app_base.wait_for_redis(sender, **kwargs)
     app_base.wait_for_db(sender, **kwargs)
     app_base.wait_for_vespa(sender, **kwargs)
+
+    # Less startup checks in multi-tenant case
+    if MULTI_TENANT:
+        return
+
     app_base.on_secondary_worker_init(sender, **kwargs)
 
 

backend/onyx/background/celery/apps/indexing.py

@@ -62,13 +62,14 @@ def on_worker_init(sender: Any, **kwargs: Any) -> None:
     SqlEngine.set_app_name(POSTGRES_CELERY_WORKER_INDEXING_APP_NAME)
     SqlEngine.init_engine(pool_size=sender.concurrency, max_overflow=sender.concurrency)
 
-    # Startup checks are not needed in multi-tenant case
-    if MULTI_TENANT:
-        return
-
     app_base.wait_for_redis(sender, **kwargs)
     app_base.wait_for_db(sender, **kwargs)
     app_base.wait_for_vespa(sender, **kwargs)
+
+    # Less startup checks in multi-tenant case
+    if MULTI_TENANT:
+        return
+
     app_base.on_secondary_worker_init(sender, **kwargs)
 
 

backend/onyx/background/celery/apps/light.py

@@ -60,13 +60,15 @@ def on_worker_init(sender: Any, **kwargs: Any) -> None:
 
     SqlEngine.set_app_name(POSTGRES_CELERY_WORKER_LIGHT_APP_NAME)
     SqlEngine.init_engine(pool_size=sender.concurrency, max_overflow=8)
-    # Startup checks are not needed in multi-tenant case
-    if MULTI_TENANT:
-        return
 
     app_base.wait_for_redis(sender, **kwargs)
     app_base.wait_for_db(sender, **kwargs)
     app_base.wait_for_vespa(sender, **kwargs)
+
+    # Less startup checks in multi-tenant case
+    if MULTI_TENANT:
+        return
+
     app_base.on_secondary_worker_init(sender, **kwargs)
 
 

backend/onyx/background/celery/apps/primary.py

@@ -84,14 +84,14 @@ def on_worker_init(sender: Any, **kwargs: Any) -> None:
     SqlEngine.set_app_name(POSTGRES_CELERY_WORKER_PRIMARY_APP_NAME)
     SqlEngine.init_engine(pool_size=8, max_overflow=0)
 
-    # Startup checks are not needed in multi-tenant case
-    if MULTI_TENANT:
-        return
-
     app_base.wait_for_redis(sender, **kwargs)
     app_base.wait_for_db(sender, **kwargs)
     app_base.wait_for_vespa(sender, **kwargs)
 
+    # Less startup checks in multi-tenant case
+    if MULTI_TENANT:
+        return
+
     logger.info("Running as the primary celery worker.")
 
     # This is singleton work that should be done on startup exactly once

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

@@ -29,7 +29,6 @@ from onyx.configs.constants import OnyxCeleryPriority
 from onyx.configs.constants import OnyxCeleryQueues
 from onyx.configs.constants import OnyxCeleryTask
 from onyx.configs.constants import OnyxRedisLocks
-from onyx.configs.constants import OnyxRedisSignals
 from onyx.db.connector import mark_ccpair_with_indexing_trigger
 from onyx.db.connector_credential_pair import fetch_connector_credential_pairs
 from onyx.db.connector_credential_pair import get_connector_credential_pair_from_id
@@ -176,7 +175,7 @@ def check_for_indexing(self: Task, *, tenant_id: str | None) -> int | None:
 
     # we need to use celery's redis client to access its redis data
     # (which lives on a different db number)
-    redis_client_celery: Redis = self.app.broker_connection().channel().client  # type: ignore
+    # redis_client_celery: Redis = self.app.broker_connection().channel().client  # type: ignore
 
     lock_beat: RedisLock = redis_client.lock(
         OnyxRedisLocks.CHECK_INDEXING_BEAT_LOCK,
@@ -319,20 +318,23 @@ def check_for_indexing(self: Task, *, tenant_id: str | None) -> int | None:
                     attempt.id, db_session, failure_reason=failure_reason
                 )
 
-        # we want to run this less frequently than the overall task
-        if not redis_client.exists(OnyxRedisSignals.VALIDATE_INDEXING_FENCES):
-            # clear any indexing fences that don't have associated celery tasks in progress
-            # tasks can be in the queue in redis, in reserved tasks (prefetched by the worker),
-            # or be currently executing
-            try:
-                task_logger.info("Validating indexing fences...")
-                validate_indexing_fences(
-                    tenant_id, self.app, redis_client, redis_client_celery, lock_beat
-                )
-            except Exception:
-                task_logger.exception("Exception while validating indexing fences")
+        # rkuo: The following code logically appears to work, but the celery inspect code may be unstable
+        # turning off for the moment to see if it helps cloud stability
 
-            redis_client.set(OnyxRedisSignals.VALIDATE_INDEXING_FENCES, 1, ex=60)
+        # we want to run this less frequently than the overall task
+        # if not redis_client.exists(OnyxRedisSignals.VALIDATE_INDEXING_FENCES):
+        #     # clear any indexing fences that don't have associated celery tasks in progress
+        #     # tasks can be in the queue in redis, in reserved tasks (prefetched by the worker),
+        #     # or be currently executing
+        #     try:
+        #         task_logger.info("Validating indexing fences...")
+        #         validate_indexing_fences(
+        #             tenant_id, self.app, redis_client, redis_client_celery, lock_beat
+        #         )
+        #     except Exception:
+        #         task_logger.exception("Exception while validating indexing fences")
+
+        #     redis_client.set(OnyxRedisSignals.VALIDATE_INDEXING_FENCES, 1, ex=60)
 
     except SoftTimeLimitExceeded:
         task_logger.info(

backend/onyx/document_index/vespa/index.py

@@ -535,7 +535,7 @@ class VespaIndex(DocumentIndex):
         if self.secondary_index_name:
             index_names.append(self.secondary_index_name)
 
-        with get_vespa_http_client() as http_client:
+        with get_vespa_http_client(http2=False) as http_client:
             for index_name in index_names:
                 params = httpx.QueryParams(
                     {
@@ -546,8 +546,12 @@ class VespaIndex(DocumentIndex):
 
                 while True:
                     try:
+                        vespa_url = (
+                            f"{DOCUMENT_ID_ENDPOINT.format(index_name=self.index_name)}"
+                        )
+                        logger.debug(f'update_single PUT on URL "{vespa_url}"')
                         resp = http_client.put(
-                            f"{DOCUMENT_ID_ENDPOINT.format(index_name=self.index_name)}",
+                            vespa_url,
                             params=params,
                             headers={"Content-Type": "application/json"},
                             json=update_dict,
@@ -619,7 +623,7 @@ class VespaIndex(DocumentIndex):
         if self.secondary_index_name:
             index_names.append(self.secondary_index_name)
 
-        with get_vespa_http_client() as http_client:
+        with get_vespa_http_client(http2=False) as http_client:
             for index_name in index_names:
                 params = httpx.QueryParams(
                     {
@@ -630,8 +634,12 @@ class VespaIndex(DocumentIndex):
 
                 while True:
                     try:
+                        vespa_url = (
+                            f"{DOCUMENT_ID_ENDPOINT.format(index_name=index_name)}"
+                        )
+                        logger.debug(f'delete_single DELETE on URL "{vespa_url}"')
                         resp = http_client.delete(
-                            f"{DOCUMENT_ID_ENDPOINT.format(index_name=index_name)}",
+                            vespa_url,
                             params=params,
                         )
                         resp.raise_for_status()

backend/onyx/document_index/vespa/shared_utils/utils.py

@@ -55,7 +55,9 @@ def remove_invalid_unicode_chars(text: str) -> str:
     return _illegal_xml_chars_RE.sub("", text)
 
 
-def get_vespa_http_client(no_timeout: bool = False) -> httpx.Client:
+def get_vespa_http_client(
+    no_timeout: bool = False, http2: bool = False
+) -> httpx.Client:
     """
     Configure and return an HTTP client for communicating with Vespa,
     including authentication if needed.
@@ -67,5 +69,5 @@ def get_vespa_http_client(no_timeout: bool = False) -> httpx.Client:
         else None,
         verify=False if not MANAGED_VESPA else True,
         timeout=None if no_timeout else VESPA_REQUEST_TIMEOUT,
-        http2=True,
+        http2=http2,
     )

backend/onyx/llm/chat_llm.py

@@ -453,7 +453,9 @@ class DefaultMultiLLM(LLM):
         if LOG_DANSWER_MODEL_INTERACTIONS:
             self.log_model_configs()
 
-        if DISABLE_LITELLM_STREAMING:
+        if (
+            DISABLE_LITELLM_STREAMING or self.config.model_name == "o1-2024-12-17"
+        ):  # TODO: remove once litellm supports streaming
             yield self.invoke(prompt, tools, tool_choice, structured_response_format)
             return
 

backend/onyx/llm/llm_provider_options.py

@@ -29,6 +29,7 @@ OPENAI_PROVIDER_NAME = "openai"
 OPEN_AI_MODEL_NAMES = [
     "o1-mini",
     "o1-preview",
+    "o1-2024-12-17",
     "gpt-4",
     "gpt-4o",
     "gpt-4o-mini",

backend/onyx/utils/telemetry.py

@@ -22,6 +22,7 @@ from onyx.utils.variable_functionality import (
 from onyx.utils.variable_functionality import noop_fallback
 from shared_configs.configs import MULTI_TENANT
 
+
 _DANSWER_TELEMETRY_ENDPOINT = "https://telemetry.onyx.app/anonymous_telemetry"
 _CACHED_UUID: str | None = None
 _CACHED_INSTANCE_DOMAIN: str | None = None
@@ -117,9 +118,12 @@ def mt_cloud_telemetry(
     event: MilestoneRecordType,
     properties: dict | None = None,
 ) -> None:
+    print(f"mt_cloud_telemetry {distinct_id} {event} {properties}")
     if not MULTI_TENANT:
+        print("mt_cloud_telemetry not MULTI_TENANT")
         return
 
+    print("mt_cloud_telemetry MULTI_TENANT")
     # MIT version should not need to include any Posthog code
     # This is only for Onyx MT Cloud, this code should also never be hit, no reason for any orgs to
     # be running the Multi Tenant version of Onyx.
@@ -137,8 +141,11 @@ def create_milestone_and_report(
     properties: dict | None,
     db_session: Session,
 ) -> None:
+    print(f"create_milestone_and_report {user} {event_type} {db_session}")
     _, is_new = create_milestone_if_not_exists(user, event_type, db_session)
+    print(f"create_milestone_and_report {is_new}")
     if is_new:
+        print("create_milestone_and_report is_new")
         mt_cloud_telemetry(
             distinct_id=distinct_id,
             event=event_type,

backend/requirements/default.txt

@@ -29,7 +29,7 @@ trafilatura==1.12.2
 langchain==0.1.17
 langchain-core==0.1.50
 langchain-text-splitters==0.0.1
-litellm==1.54.1
+litellm==1.55.4
 lxml==5.3.0
 lxml_html_clean==0.2.2
 llama-index==0.9.45

backend/requirements/model_server.txt

@@ -12,5 +12,5 @@ torch==2.2.0
 transformers==4.39.2
 uvicorn==0.21.1
 voyageai==0.2.3
-litellm==1.54.1
+litellm==1.55.4
 sentry-sdk[fastapi,celery,starlette]==2.14.0

web/public/Microsoft.svg

@@ -0,0 +1,6 @@
+<svg width="16" height="16" viewBox="0 0 16 16" fill="none" xmlns="http://www.w3.org/2000/svg">
+<rect x="1.33325" y="1.3335" width="6.33333" height="6.33333" fill="#F25022"/>
+<rect x="8.33325" y="1.3335" width="6.33333" height="6.33333" fill="#80BA01"/>
+<rect x="8.33325" y="8.3335" width="6.33333" height="6.33333" fill="#FFB902"/>
+<rect x="1.33325" y="8.3335" width="6.33333" height="6.33333" fill="#02A4EF"/>
+</svg>

web/public/Mistral.svg

@@ -0,0 +1 @@
+<svg viewBox="0 0 512 512" xmlns="http://www.w3.org/2000/svg" fill-rule="evenodd" clip-rule="evenodd" stroke-linejoin="round" stroke-miterlimit="2"><path d="M189.08 303.228H94.587l.044-94.446h94.497l-.048 94.446z" fill="#1c1c1b" fill-rule="nonzero"/><path d="M283.528 397.674h-94.493l.044-94.446h94.496l-.047 94.446z" fill="#1c1c1b" fill-rule="nonzero"/><path d="M283.575 303.228H189.08l.046-94.446h94.496l-.047 94.446z" fill="#1c1c1b" fill-rule="nonzero"/><path d="M378.07 303.228h-94.495l.044-94.446h94.498l-.047 94.446zM189.128 208.779H94.633l.044-94.448h94.498l-.047 94.448zM378.115 208.779h-94.494l.045-94.448h94.496l-.047 94.448zM94.587 303.227H.093l.044-96.017h94.496l-.046 96.017z" fill="#1c1c1b" fill-rule="nonzero"/><path d="M94.633 208.779H.138l.046-94.448H94.68l-.047 94.448z" fill="#1c1c1b" fill-rule="nonzero"/><path d="M94.68 115.902H.185L.23 19.885h94.498l-.047 96.017zM472.657 114.331h-94.495l.044-94.446h94.497l-.046 94.446zM94.54 399.244H.046l.044-97.588h94.497l-.047 97.588z" fill="#1c1c1b" fill-rule="nonzero"/><path d="M94.495 492.123H0l.044-94.446H94.54l-.045 94.446zM472.563 303.228H378.07l.044-94.446h94.496l-.047 94.446zM472.61 208.779h-94.495l.044-94.448h94.498l-.047 94.448z" fill="#1c1c1b" fill-rule="nonzero"/><path d="M472.517 397.674h-94.494l.044-94.446h94.497l-.047 94.446z" fill="#1c1c1b" fill-rule="nonzero"/><path d="M472.47 492.121h-94.493l.044-96.017h94.496l-.047 96.017z" fill="#1c1c1b" fill-rule="nonzero"/><path d="M228.375 303.22h-96.061l.046-94.446h96.067l-.052 94.446z" fill="#ff7000" fill-rule="nonzero"/><path d="M322.827 397.666h-94.495l.044-96.018h94.498l-.047 96.018z" fill="#ff4900" fill-rule="nonzero"/><path d="M324.444 303.22h-97.636l.046-94.446h97.638l-.048 94.446z" fill="#ff7000" fill-rule="nonzero"/><path d="M418.938 303.22h-96.064l.045-94.446h96.066l-.047 94.446z" fill="#ff7000" fill-rule="nonzero"/><path d="M228.423 208.77H132.36l.045-94.445h96.066l-.05 94.446zM418.985 208.77H322.92l.044-94.445h96.069l-.048 94.446z" fill="#ffa300" fill-rule="nonzero"/><path d="M133.883 304.79H39.392l.044-96.017h94.496l-.049 96.017z" fill="#ff7000" fill-rule="nonzero"/><path d="M133.929 208.77H39.437l.044-95.445h94.496l-.048 95.445z" fill="#ffa300" fill-rule="nonzero"/><path d="M133.976 114.325H39.484l.044-94.448h94.497l-.05 94.448zM511.954 115.325h-94.493l.044-95.448h94.497l-.048 95.448z" fill="#ffce00" fill-rule="nonzero"/><path d="M133.836 399.667H39.345l.044-96.447h94.496l-.049 96.447z" fill="#ff4900" fill-rule="nonzero"/><path d="M133.79 492.117H39.3l.044-94.448h94.496l-.049 94.448z" fill="#ff0107" fill-rule="nonzero"/><path d="M511.862 303.22h-94.495l.046-94.446h94.496l-.047 94.446z" fill="#ff7000" fill-rule="nonzero"/><path d="M511.907 208.77h-94.493l.044-94.445h94.496l-.047 94.446z" fill="#ffa300" fill-rule="nonzero"/><path d="M511.815 398.666h-94.493l.044-95.447h94.496l-.047 95.447z" fill="#ff4900" fill-rule="nonzero"/><path d="M511.77 492.117h-94.496l.046-94.448h94.496l-.047 94.448z" fill="#ff0107" fill-rule="nonzero"/></svg>

web/src/app/admin/configuration/llm/interfaces.ts

@@ -1,8 +1,12 @@
 import {
   AnthropicIcon,
+  AmazonIcon,
   AWSIcon,
   AzureIcon,
   CPUIcon,
+  MicrosoftIconSVG,
+  MistralIcon,
+  MetaIcon,
   OpenAIIcon,
   GeminiIcon,
   OpenSourceIcon,
@@ -72,12 +76,25 @@ export const getProviderIcon = (providerName: string, modelName?: string) => {
   switch (providerName) {
     case "openai":
       // Special cases for openai based on modelName
+      if (modelName?.toLowerCase().includes("amazon")) {
+        return AmazonIcon;
+      }
+      if (modelName?.toLowerCase().includes("phi")) {
+        return MicrosoftIconSVG;
+      }    
+      if (modelName?.toLowerCase().includes("mistral")) {
+        return MistralIcon;
+      }    
+      if (modelName?.toLowerCase().includes("llama")) {
+        return MetaIcon;
+      }      
       if (modelName?.toLowerCase().includes("gemini")) {
         return GeminiIcon;
       }
       if (modelName?.toLowerCase().includes("claude")) {
         return AnthropicIcon;
       }
+
       return OpenAIIcon; // Default for openai
     case "anthropic":
       return AnthropicIcon;

web/src/components/icons/icons.tsx

@@ -39,7 +39,10 @@ import Image, { StaticImageData } from "next/image";
 import jiraSVG from "../../../public/Jira.svg";
 import confluenceSVG from "../../../public/Confluence.svg";
 import openAISVG from "../../../public/Openai.svg";
+import amazonSVG from "../../../public/Amazon.svg";
 import geminiSVG from "../../../public/Gemini.svg";
+import metaSVG from "../../../public/Meta.svg";
+import mistralSVG from "../../../public/Mistral.svg";
 import openSourceIcon from "../../../public/OpenSource.png";
 import litellmIcon from "../../../public/LiteLLM.jpg";
 
@@ -49,6 +52,7 @@ import asanaIcon from "../../../public/Asana.png";
 import anthropicSVG from "../../../public/Anthropic.svg";
 import nomicSVG from "../../../public/nomic.svg";
 import microsoftIcon from "../../../public/microsoft.png";
+import microsoftSVG from "../../../public/Microsoft.svg";
 import mixedBreadSVG from "../../../public/Mixedbread.png";
 
 import OCIStorageSVG from "../../../public/OCI.svg";
@@ -1104,6 +1108,26 @@ export const GeminiIcon = ({
   className = defaultTailwindCSS,
 }: IconProps) => <LogoIcon size={size} className={className} src={geminiSVG} />;
 
+export const AmazonIcon = ({
+  size = 16,
+  className = defaultTailwindCSS,
+}: IconProps) => <LogoIcon size={size} className={className} src={amazonSVG} />;
+
+export const MetaIcon = ({
+  size = 16,
+  className = defaultTailwindCSS,
+}: IconProps) => <LogoIcon size={size} className={className} src={metaSVG} />;
+
+export const MicrosoftIconSVG = ({
+  size = 16,
+  className = defaultTailwindCSS,
+}: IconProps) => <LogoIcon size={size} className={className} src={microsoftSVG} />;
+
+export const MistralIcon = ({
+  size = 16,
+  className = defaultTailwindCSS,
+}: IconProps) => <LogoIcon size={size} className={className} src={mistralSVG} />;
+
 export const VoyageIcon = ({
   size = 16,
   className = defaultTailwindCSS,

web/src/lib/hooks.ts

@@ -299,6 +299,7 @@ const MODEL_DISPLAY_NAMES: { [key: string]: string } = {
   // OpenAI models
   "o1-mini": "O1 Mini",
   "o1-preview": "O1 Preview",
+  "o1-2024-12-17": "O1",
   "gpt-4": "GPT 4",
   "gpt-4o": "GPT 4o",
   "gpt-4o-2024-08-06": "GPT 4o (Structured Outputs)",
@@ -318,6 +319,21 @@ const MODEL_DISPLAY_NAMES: { [key: string]: string } = {
   "gpt-3.5-turbo-16k-0613": "GPT 3.5 Turbo 16k (June 2023)",
   "gpt-3.5-turbo-0301": "GPT 3.5 Turbo (March 2023)",
 
+  // Amazon models
+  "amazon.nova-micro@v1": "Amazon Nova Micro",
+  "amazon.nova-lite@v1": "Amazon Nova Lite",
+  "amazon.nova-pro@v1": "Amazon Nova Pro",
+
+  // Meta models
+  "llama-3.2-90b-vision-instruct": "Llama 3.2 90B",
+  "llama-3.2-11b-vision-instruct": "Llama 3.2 11B",
+  "llama-3.3-70b-instruct": "Llama 3.3 70B",
+
+  // Microsoft models
+  "phi-3.5-mini-instruct": "Phi 3.5 Mini",
+  "phi-3.5-moe-instruct": "Phi 3.5 MoE",
+  "phi-3.5-vision-instruct": "Phi 3.5 Vision",
+
   // Anthropic models
   "claude-3-opus-20240229": "Claude 3 Opus",
   "claude-3-sonnet-20240229": "Claude 3 Sonnet",
@@ -329,6 +345,9 @@ const MODEL_DISPLAY_NAMES: { [key: string]: string } = {
   "claude-3-5-sonnet-20241022": "Claude 3.5 Sonnet (New)",
   "claude-3-5-sonnet-v2@20241022": "Claude 3.5 Sonnet (New)",
   "claude-3.5-sonnet-v2@20241022": "Claude 3.5 Sonnet (New)",
+  "claude-3-5-haiku-20241022": "Claude 3.5 Haiku",
+  "claude-3-5-haiku@20241022": "Claude 3.5 Haiku",
+  "claude-3.5-haiku@20241022": "Claude 3.5 Haiku",
 
   // Google Models
   "gemini-1.5-pro": "Gemini 1.5 Pro",
@@ -337,6 +356,11 @@ const MODEL_DISPLAY_NAMES: { [key: string]: string } = {
   "gemini-1.5-flash-001": "Gemini 1.5 Flash",
   "gemini-1.5-pro-002": "Gemini 1.5 Pro (v2)",
   "gemini-1.5-flash-002": "Gemini 1.5 Flash (v2)",
+  "gemini-2.0-flash-exp": "Gemini 2.0 Flash (Experimental)",
+
+  // Mistral Models
+  "mistral-large-2411": "Mistral Large 24.11",
+  "mistral-large@2411": "Mistral Large 24.11",
 
   // Bedrock models
   "meta.llama3-1-70b-instruct-v1:0": "Llama 3.1 70B",

web/src/lib/llm/utils.ts

@@ -74,6 +74,8 @@ const MODEL_NAMES_SUPPORTING_IMAGE_INPUT = [
   "claude-3-opus-20240229",
   "claude-3-sonnet-20240229",
   "claude-3-haiku-20240307",
+  // custom claude names
+  "claude-3.5-sonnet-v2@20241022",
   // claude names with AWS Bedrock Suffix
   "claude-3-opus-20240229-v1:0",
   "claude-3-sonnet-20240229-v1:0",
@@ -93,6 +95,13 @@ const MODEL_NAMES_SUPPORTING_IMAGE_INPUT = [
   "gemini-1.5-flash-001",
   "gemini-1.5-pro-002",
   "gemini-1.5-flash-002",
+  "gemini-2.0-flash-exp",
+  // amazon models
+  "amazon.nova-lite@v1",
+  "amazon.nova-pro@v1",
+  // meta models
+  "llama-3.2-90b-vision-instruct",
+  "llama-3.2-11b-vision-instruct"  
 ];
 
 export function checkLLMSupportsImageInput(model: string) {