add support for openwebui

arch/envoy.template.yaml

@@ -728,7 +728,7 @@ static_resources:
               - endpoint:
                   address:
                     socket_address:
-                      address: 0.0.0.0
+                      address: host.docker.internal
                       port_value: 9091
                   hostname: localhost
 

arch/tools/cli/docker_cli.py

@@ -52,6 +52,7 @@ def docker_start_archgw_detached(
     port_mappings = [
         f"{prompt_gateway_port}:{prompt_gateway_port}",
         f"{llm_gateway_port}:{llm_gateway_port}",
+        f"{llm_gateway_port+1}:{llm_gateway_port+1}",
         "19901:9901",
     ]
     port_mappings_args = [item for port in port_mappings for item in ("-p", port)]

crates/brightstaff/src/handlers/mod.rs

@@ -1 +1,2 @@
 pub mod chat_completions;
+pub mod models;

crates/brightstaff/src/main.rs

@@ -1,4 +1,5 @@
 use brightstaff::handlers::chat_completions::chat_completions;
+use brightstaff::handlers::models::list_models;
 use brightstaff::router::llm_router::RouterService;
 use bytes::Bytes;
 use common::configuration::Configuration;
@@ -72,6 +73,8 @@ async fn main() -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
 
     let arch_config = Arc::new(config);
 
+    let llm_providers = Arc::new(arch_config.llm_providers.clone());
+
     debug!(
         "arch_config: {:?}",
         &serde_json::to_string(arch_config.as_ref()).unwrap()
@@ -104,10 +107,12 @@ async fn main() -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
         let router_service = Arc::clone(&router_service);
         let llm_provider_endpoint = llm_provider_endpoint.clone();
 
+        let llm_providers = llm_providers.clone();
         let service = service_fn(move |req| {
             let router_service = Arc::clone(&router_service);
             let parent_cx = extract_context_from_request(&req);
             let llm_provider_endpoint = llm_provider_endpoint.clone();
+            let llm_providers = llm_providers.clone();
 
             async move {
                 match (req.method(), req.uri().path()) {
@@ -116,6 +121,35 @@ async fn main() -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
                             .with_context(parent_cx)
                             .await
                     }
+                    (&Method::GET, "/v1/models") => {
+                        Ok(list_models(llm_providers).await)
+                    }
+                    (&Method::OPTIONS, "/v1/models") => {
+                        let mut response = Response::new(empty());
+                        *response.status_mut() = StatusCode::NO_CONTENT;
+                        response.headers_mut().insert(
+                            "Allow",
+                            "GET, OPTIONS".parse().unwrap(),
+                        );
+                        response.headers_mut().insert(
+                            "Access-Control-Allow-Origin",
+                            "*".parse().unwrap(),
+                        );
+                        response.headers_mut().insert(
+                            "Access-Control-Allow-Headers",
+                            "Authorization, Content-Type".parse().unwrap(),
+                        );
+                        response.headers_mut().insert(
+                            "Access-Control-Allow-Methods",
+                            "GET, POST, OPTIONS".parse().unwrap(),
+                        );
+                        response.headers_mut().insert(
+                            "Content-Type",
+                            "application/json".parse().unwrap(),
+                        );
+
+                        Ok(response)
+                    }
                     _ => {
                         let mut not_found = Response::new(empty());
                         *not_found.status_mut() = StatusCode::NOT_FOUND;

crates/common/src/api/open_ai.rs

@@ -1,7 +1,10 @@
-use crate::consts::{ARCH_FC_MODEL_NAME, ASSISTANT_ROLE};
+use crate::{
+    configuration::LlmProvider,
+    consts::{ARCH_FC_MODEL_NAME, ASSISTANT_ROLE},
+};
+use core::{panic, str};
 use serde::{ser::SerializeMap, Deserialize, Serialize};
 use serde_yaml::Value;
-use core::panic;
 use std::{
     collections::{HashMap, VecDeque},
     fmt::Display,
@@ -420,6 +423,45 @@ pub fn to_server_events(chunks: Vec<ChatCompletionStreamResponse>) -> String {
     response_str
 }
 
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct ModelDetail {
+    pub id: String,
+    pub object: String,
+    pub created: usize,
+    pub owned_by: String,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub enum ModelObject {
+    #[serde(rename = "list")]
+    List,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct Models {
+    pub object: ModelObject,
+    pub data: Vec<ModelDetail>,
+}
+
+impl From<Vec<LlmProvider>> for Models {
+    fn from(llm_providers: Vec<LlmProvider>) -> Self {
+        let data = llm_providers
+            .iter()
+            .map(|provider| ModelDetail {
+                id: provider.model.as_ref().unwrap().clone(),
+                object: "model".to_string(),
+                created: 1721172741,
+                owned_by: "system".to_string(),
+            })
+            .collect();
+
+        Models {
+            object: ModelObject::List,
+            data,
+        }
+    }
+}
+
 #[cfg(test)]
 mod test {
     use crate::api::open_ai::{ChatCompletionsRequest, ContentType, MultiPartContentType};
@@ -775,7 +817,10 @@ data: [DONE]
         if let Some(ContentType::MultiPart(multi_part_content)) =
             chat_completions_request.messages[0].content.as_ref()
         {
-            assert_eq!(multi_part_content[0].content_type, MultiPartContentType::Text);
+            assert_eq!(
+                multi_part_content[0].content_type,
+                MultiPartContentType::Text
+            );
             assert_eq!(
                 multi_part_content[0].text,
                 Some("What city do you want to know the weather for?".to_string())
@@ -815,22 +860,24 @@ data: [DONE]
             chat_completions_request.messages[0].content.as_ref()
         {
             assert_eq!(multi_part_content.len(), 2);
-            assert_eq!(multi_part_content[0].content_type, MultiPartContentType::Text);
+            assert_eq!(
+                multi_part_content[0].content_type,
+                MultiPartContentType::Text
+            );
             assert_eq!(
                 multi_part_content[0].text,
                 Some("What city do you want to know the weather for?".to_string())
             );
-            assert_eq!(multi_part_content[1].content_type, MultiPartContentType::Text);
             assert_eq!(
-                multi_part_content[1].text,
+                multi_part_content[1].content_type,
-                Some("hello world".to_string())
+                MultiPartContentType::Text
             );
+            assert_eq!(multi_part_content[1].text, Some("hello world".to_string()));
         } else {
             panic!("Expected MultiPartContent");
         }
     }
 
-
     #[test]
     fn stream_chunk_parse_claude() {
         const CHUNK_RESPONSE: &str = r#"data: {"id":"msg_01DZDMxYSgq8aPQxMQoBv6Kb","choices":[{"index":0,"delta":{"role":"assistant"}}],"created":1747685264,"model":"claude-3-7-sonnet-latest","object":"chat.completion.chunk"}

demos/use_cases/llm_routing/docker-compose.yaml

@@ -1,17 +1,15 @@
 services:
 
-  chatbot_ui:
+
-    build:
+  open-web-ui:
-      context: ../../shared/chatbot_ui
+    image: ghcr.io/open-webui/open-webui:main
-      dockerfile: Dockerfile
+    container_name: open-webui
+    restart: always
     ports:
-      - "18080:8080"
+      - "8080:8080"
     environment:
-      - CHAT_COMPLETION_ENDPOINT=http://host.docker.internal:12000/v1
+      - DEFAULT_MODEL=gpt-4o-mini
-    extra_hosts:
+      - OPENAI_API_KEY=your_secret_key
-      - "host.docker.internal:host-gateway"
-    volumes:
-      - ./arch_config.yaml:/app/arch_config.yaml
 
   jaeger:
     build: