RAG Service
Unified Test
Toggle Theme
Unified RAG Test Interface
Model Selection
Choose Model (per-request override)
Override ACTIVE
Model:
—
Model alias
Loading models…
All
Azure
OpenAI
OpenRouter
Anthropic
Gemini
Mistral
Multimodal
Reasoning
Show hidden
Tip: Type to filter. Click an option to apply the override. Use the toggle to reveal models hidden by validation.
Auto-applies to requests from this page. Defaults come from settings unless overridden.
💾 Save Local
🌐 Save Server Default
Clear Override
Deep check
Recheck
Quick Start:
1) Type your question (or leave the example) 2) (Optional) adjust Top K or indexes 3) Press
Ctrl
Enter
(or click Send) 4) Review Answer & Sources → provide feedback.
Loading prompts metadata…
Question & Chat
What can you tell me about yourself?
0 chars
💬
Chat Mode
OFF
Ctrl/Cmd+Enter to send
Conversation History
Reset
Streaming
🌊 Enable Streaming
💬 Stream tokens as they're generated for faster perceived response (real-time output).
Multimodal (optional)
Use multimodal variant (
/api/v2/flexible-rag-mm
) and include images
Drag & drop image files here, or add by URL below
Add image
Tip: You can paste HTTPS URLs or drop image files (converted to data URLs client-side).
Basic Retrieval Settings
Retrieval (basic)
Enable Knowledge Base Lookup
📚 When enabled, searches your indexed documents for relevant information (usually takes about 6 seconds)
Search Query
(defaults to Question)
Top K
Vector Search
Yes
No
Indexes
(choose one, multiple, or All)
Select All
Clear
LLM Parameters
Defaults
Loading defaults…
LLM defaults not available for this environment.
Temperature
(0.0-2.0)
i
Top P
(0.0-1.0)
i
Max Tokens
i
Frequency Penalty
i
Presence Penalty
i
N (completions)
i
Reasoning Effort
i
None / Auto
Low
Medium
High
Save Parameters
Templating & Overrides (advanced)
Templating
System Prompt Source
Default
Inline
File (blob / fs)
Response Template Source
Default
Inline
File (blob / fs)
View Current Default System Prompt
(loading...)
View Current Default Response Template
(loading...)
System Prompt File Path
Choose from available (fs/blob)
Loading...
Preview Selected System Prompt
(none selected)
Inline System Prompt (Jinja2)
You are a helpful AI assistant. Current question:
Response Template File Path
Choose from available (fs/blob)
Loading...
Preview Selected Response Template
(none selected)
Inline Response Template (Jinja2)
## Answer
Template Variables
+ Add Variable
Advanced JSON Overrides
Metadata (JSON)
{"user_id":"demo_user"}
Example: {"user_id":"abc","session":"123"}
Config Overrides (JSON)
{}
Leave {} unless you need to override internal config values.
Send Request
Clear
Response
Was this helpful?
Yes
No
Streaming output
Answer
Extracted documents
Metadata
Raw JSON
Toggle raw response