Skip to main content
POST
/
ai
/
v2
/
chat
/
completions
/
grounded
curl --request POST \
  --url https://platform.ai.gloo.com/ai/v2/chat/completions/grounded \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "messages": [
    {
      "role": "user",
      "content": "What are practical ways to build stronger community in a local church?"
    }
  ],
  "auto_routing": true,
  "sources_limit": 3,
  "tradition": "evangelical"
}
'
{
  "id": "gen-1769792268-xTKWyq7A9HmEhxzDsAJj",
  "object": "chat.completion",
  "created": 1769792268,
  "model": "gloo-anthropic-claude-sonnet-4.5",
  "provider": "Gloo AI",
  "model_family": "Anthropic",
  "auto_routing": true,
  "routing_mechanism": "auto_routing",
  "routing_tier": "tier_3",
  "routing_confidence": 0.782,
  "tradition": "evangelical",
  "sources_returned": true,
  "citations": [
    {
      "item_title": "Daily Prayer",
      "item_url": "https://www.studio.ai.gloo.com",
      "author": [
        "John Doe"
      ],
      "publisher": "The Publisher",
      "publication_date": "Apr 08 2017",
      "snippets": [
        "Daily prayer is an essential part of life as...",
        "Making time to spend with the Lord is..."
      ]
    }
  ],
  "choices": [
    {
      "finish_reason": "stop",
      "index": 0,
      "message": {
        "content": "Based on the retrieved sources, here are practical ways to build stronger community...",
        "role": "assistant"
      }
    }
  ],
  "usage": {
    "completion_tokens": 592,
    "prompt_tokens": 2137,
    "total_tokens": 2729
  }
}

Documentation Index

Fetch the complete documentation index at: https://docs.gloo.com/llms.txt

Use this file to discover all available pages before exploring further.

Error Reference: Grounded Completions

Common errors: retrieval/source failures, no relevant content retrieved, plus all Completions V2 errors.

Authorizations

​
Authorization
string
header
required

Bearer authentication header of the form Bearer <token>, where <token> is your auth token.

Body

application/json

Request body for the grounded completions endpoint. Exactly one routing mechanism (auto_routing, model, or model_family) must be specified. Messages are optional when omitted and default to a single user message of "Hello world!". An explicit empty messages array is invalid and returns 400.

​
messages
LlmMessage · object[]

Chat message history with role and content fields. If omitted, the current service defaults to one user message with content "Hello world!". Explicitly passing an empty array is invalid.

​
auto_routing
boolean

Enables intelligent model selection. Mutually exclusive with model and model_family.

​
model
string

Specific Gloo model identifier. Mutually exclusive with auto_routing and model_family.

​
model_family
enum<string>

Provider family for model selection. Mutually exclusive with auto_routing and model.

Available options:
openai,
anthropic,
google,
open source
​
rag_publisher
string
default:GlooGrounded

Publisher name to retrieve sources from. Defaults to GlooGrounded when omitted.

​
sources_limit
integer
default:3

Number of sources to retrieve for grounding (1-10).

Required range: 1 <= x <= 10
​
tradition
enum<string>

Theological perspective to apply. Options: evangelical, catholic, mainline, not_faith_specific.

Available options:
evangelical,
catholic,
mainline,
not_faith_specific
​
stream
boolean
default:false

Enables streaming responses via server-sent events.

​
temperature
number
default:0.7

Sampling temperature controlling randomness.

Required range: 0 <= x <= 2
​
max_tokens
integer

Maximum number of tokens to generate in the response.

Required range: x >= 1
​
tools
Tool · object[]

Function calling definitions.

​
tool_choice
default:none

Controls which tool (if any) the model should use.

Available options:
none,
auto,
required
​
parallel_tool_calls
boolean
default:true

Allow parallel tool execution.

​
include_citations
boolean
default:false

Include citation metadata for sources utilized by RAG. For streaming responses, citations are prepended as the first SSE event before any content chunks.

​
stream_options
Stream Options · object

Streaming options. Use include_usage=true with stream=true to receive a final usage event.

Response

Successful grounded completion response. If stream is false or omitted, the response is JSON. If stream=true, the response is a Server-Sent Events stream. Once streaming has started, failures are delivered as stream events rather than non-2xx HTTP responses. Content moderation ends the stream with a content_filter event. When include_citations=true and citations are available, the first stream event contains citations.

Response from the grounded completions endpoint.

​
id
string

Unique completion identifier.

​
object
string
default:chat.completion

Object type, always 'chat.completion'.

​
created
integer

Unix timestamp of when the completion was created.

​
model
string

The model that was selected and used for the completion.

​
provider
string

The model provider name.

​
model_family
string

Provider family of the selected model.

​
auto_routing
boolean

Indicates whether auto routing was used.

​
routing_mechanism
string

The routing method used: auto_routing, model_family, or direct_model_selection.

​
routing_tier
string

The performance tier assigned by the routing system.

​
routing_confidence
number

Confidence score for the routing decision (0.0 to 1.0).

​
tradition
string

The theological perspective that was applied to the response.

​
sources_returned
boolean

Whether RAG was used to retrieve sources and ground the response.

​
citations
CitationObject · object[] | null

Array of source citations used to ground the response. Only present when include_citations is true in the request.

​
choices
CompletionsV2Choice · object[]

List of completion choices.

​
usage
CompletionsV2Usage · object

Token consumption metrics.

​
service_tier
null
​
system_fingerprint
null