Generate an AI response grounded in your documents. Retrieves relevant chunks from the specified vector stores and uses them as context for the LLM (Retrieval-Augmented Generation).
curl Python
curl -X POST https://api.upstage.ai/v2/responses \ -H "Authorization: Bearer $UPSTAGE_API_KEY " \ -H "Content-Type: application/json" \ -d '{ "model": "solar-pro4", "input": "What are the key features?", "tools": [ { "type": "file_search", "vector_store_ids": ["vs_abc123def456"], "max_num_results": 5 } ] }'
{ "id" : "resp_abc123" , "object" : "response" , "created_at" : 1711645000 , "model" : "solar-pro4" , "output" : [ { "type" : "file_search_call" , "id" : "fs_call_001" , "status" : "completed" , "queries" : [ "What are the key features?" ], "results" : [ { "file_id" : "file_xyz789" , "filename" : "features.pdf" , "text" : "The key features include..." , "score" : 0.95 , "attributes" : {} } ] }, { "type" : "message" , "id" : "msg_001" , "role" : "assistant" , "content" : [ { "type" : "output_text" , "text" : "Based on the documents, the key features are..." , "annotations" : [] } ] } ] } Set stream=True to receive Server-Sent Events (SSE) in real-time.
curl Python
curl -X POST https://api.upstage.ai/v2/responses \ -H "Authorization: Bearer $UPSTAGE_API_KEY " \ -H "Content-Type: application/json" \ -d '{ "model": "solar-pro4", "input": "Summarize the document.", "tools": [ { "type": "file_search", "vector_store_ids": ["vs_abc123def456"] } ], "stream": true }'
Event Description response.createdResponse object created. file_search_call.searchingFile search is in progress. file_search_call.completedFile search completed with results. output_text.deltaIncremental text chunk from the LLM. output_text.doneLLM text generation completed. response.completedFull response completed. [DONE]Final signal indicating the stream has ended.