Merge pull request #276 from inclusionAI/feature/api-server

Feature/api server

Author: ZhuangCY

examples/gaia/cmd/agent_deploy/gaia_agent/.env.template

@@ -1,13 +1,13 @@
 # LLM Model Config
-LLM_PROVIDER = {YOUR_CONFIG:ant/openai}
+# LLM_PROVIDER = {YOUR_CONFIG}
 LLM_MODEL_NAME = {YOUR_CONFIG}
 LLM_API_KEY = {YOUR_CONFIG}
 LLM_BASE_URL = {YOUR_CONFIG}
 LLM_TEMPERATURE = 0.0
 
 # ===============Path Configurations=================
 # GAIA_DATASET_PATH="/path/to/your/gaia-benchmark/GAIA/2023"
-# AWORLD_WORKSPACE="/path/to/your/workspace_dir"
+AWORLD_WORKSPACE="/tmp"
 
 # ===============MCP Server Configurations=================
 # [Google Search API](https://developers.google.com/custom-search/v1/introduction)
@@ -29,5 +29,21 @@ IMAGE_LLM_API_KEY={YOUR_CONFIG}
 IMAGE_LLM_BASE_URL=https://openrouter.ai/api/v1
 IMAGE_LLM_MODEL_NAME=anthropic/claude-3.7-sonnet
 
+# Video Server
+VIDEO_LLM_API_KEY={YOUR_CONFIG}
+VIDEO_LLM_BASE_URL=https://openrouter.ai/api/v1
+VIDEO_LLM_MODEL_NAME=gpt-4o
+VIDEO_LLM_TEMPERATURE=1.0
+
+# Code Server
+CODE_LLM_API_KEY={YOUR_CONFIG}
+CODE_LLM_BASE_URL=https://openrouter.ai/api/v1
+CODE_LLM_MODEL_NAME=anthropic/claude-sonnet-4
+
+# Think Server
+THINK_LLM_API_KEY={YOUR_CONFIG}
+THINK_LLM_BASE_URL=https://openrouter.ai/api/v1
+THINK_LLM_MODEL_NAME=deepseek/deepseek-r1-0528:free
+
 # [E2B Server](https://e2b.dev/docs/quickstart)
 E2B_API_KEY={YOUR_CONFIG}

examples/gaia/mcp_collections/intelligence/code.py

@@ -48,9 +48,9 @@ def __init__(self, arguments: ActionArguments) -> None:
         # Initialize code generation model configuration
         self._llm_config = AgentConfig(
             llm_provider="openai",
-            llm_model_name="anthropic/claude-sonnet-4",
-            llm_api_key=os.getenv("LLM_API_KEY"),
-            llm_base_url=os.getenv("LLM_BASE_URL"),
+            llm_model_name=os.getenv("CODE_LLM_MODEL_NAME", "anthropic/claude-sonnet-4"),
+            llm_api_key=os.getenv("CODE_LLM_API_KEY"),
+            llm_base_url=os.getenv("CODE_LLM_BASE_URL"),
         )
 
         self._color_log("Code Generation Service initialized", Color.green, "debug")

examples/gaia/mcp_collections/intelligence/think.py

@@ -31,9 +31,9 @@ def __init__(self, arguments: ActionArguments) -> None:
         self._llm_config = AgentConfig(
             llm_provider="openai",
             # llm_model_name="google/gemini-2.5-flash-preview-05-20:thinking",
-            llm_model_name="deepseek/deepseek-r1-0528:free",
-            llm_api_key=os.getenv("LLM_API_KEY"),
-            llm_base_url=os.getenv("LLM_BASE_URL"),
+            llm_model_name=os.getenv("THINK_LLM_MODEL_NAME", "deepseek/deepseek-r1-0528:free"),
+            llm_api_key=os.getenv("THINK_LLM_API_KEY"),
+            llm_base_url=os.getenv("THINK_LLM_BASE_URL"),
         )
 
         self._color_log("Intelligence Reasoning Service initialized", Color.green, "debug")

examples/gaia/mcp_collections/media/video.py

@@ -371,13 +371,13 @@ def _analyze_frame_chunk(self, chunk_data: tuple[int, list, str]) -> tuple[int,
                 get_llm_model(
                     conf=AgentConfig(
                         llm_provider="openai",
-                        llm_model_name=os.getenv("LLM_MODEL_NAME", "gpt-4o"),
-                        llm_api_key=os.getenv("LLM_API_KEY"),
-                        llm_base_url=os.getenv("LLM_BASE_URL"),
+                        llm_model_name=os.getenv("VIDEO_LLM_MODEL_NAME"),
+                        llm_api_key=os.getenv("VIDEO_LLM_API_KEY"),
+                        llm_base_url=os.getenv("VIDEO_LLM_BASE_URL"),
                     )
                 ),
                 inputs,
-                temperature=float(os.getenv("LLM_TEMPERATURE", "1.0")),
+                temperature=float(os.getenv("VIDEO_LLM_TEMPERATURE", "1.0")),
             )
             analysis_result = response.content
             self._color_log(f"✅ Completed analysis for chunk {chunk_index + 1}", Color.green)
@@ -579,13 +579,13 @@ async def mcp_summarize_video(
                         get_llm_model(
                             conf=AgentConfig(
                                 llm_provider="openai",
-                                llm_model_name=os.getenv("LLM_MODEL_NAME", "gpt-4o"),
-                                llm_api_key=os.getenv("LLM_API_KEY", "your_openai_api_key"),
-                                llm_base_url=os.getenv("LLM_BASE_URL", "your_openai_base_url"),
+                                llm_model_name=os.getenv("VIDEO_LLM_MODEL_NAME", "gpt-4o"),
+                                llm_api_key=os.getenv("VIDEO_LLM_API_KEY"),
+                                llm_base_url=os.getenv("VIDEO_LLM_BASE_URL"),
                             )
                         ),
                         inputs,
-                        temperature=float(os.getenv("LLM_TEMPERATURE", "1.0")),
+                        temperature=float(os.getenv("VIDEO_LLM_TEMPERATURE", "1.0")),
                     )
                     cur_summary = response.content
                 except Exception as e: