PageIndex

🔧 PageIndexClient

PageIndexClient has two independent sides, each either local or cloud-managed:

index — where documents live and how they are processed. With a PageIndex API key they live in your cloud account and are indexed by the managed pipeline. Without one, they are indexed on your machine by the open-source pipeline (using your own LLM provider key) and stored under storage_path.
chat — who answers. With a chat model configured, the document-QA agent runs in your process against your own model and credentials — in both index modes. On a cloud client with no chat model, PageIndex’s managed chat answers.

An API key moves your documents, never your model. chat_model always means your own model on your own keys.


from pageindex import PageIndexClient
 
# Local documents + your model
client = PageIndexClient(index="gpt-5.6-luna", chat="gpt-5.6-sol")
 
# Cloud documents + your model
client = PageIndexClient(index="cloud", chat="gpt-5.6-sol")

index="cloud" (or "pageindex-cloud") is the keyless cloud spelling — the key is read from the PAGEINDEX_API_KEY environment variable. That variable is only consulted when your code explicitly says cloud: a bare PageIndexClient() stays local regardless of the environment.

Constructor parameters

Name	Type	Description	Default
api_key	`string`	PageIndex cloud API key . Omit for local mode.	None
index	`string` or `dict`	The index side, grouped: `"cloud"`, `"local"`, a local index model name, or a dict (`{"api_key"}` for cloud; `{"model", "summary_model", "backend", "storage_path"}` for local).	None
chat	`string` or `dict`	The chat side, grouped: a model name (your own model), `"cloud"` (managed chat, cloud clients only), `"local"`, or `{"model", "backend"}`.	None
mode	`string`	Cross-check on where documents live — `"cloud"` or `"local"`. Always optional; the other arguments already carry the mode.	None
index_model	`string`	Local only — model used to index documents (structure and summaries).	`gpt-5.6-luna`
chat_model	`string`	Your own model for the chat surfaces. On a cloud client, setting it runs the agent in your process over the cloud documents.	`gpt-5.6-sol`
model	`string`	Local only — one model for both roles. The role-specific arguments win over it.	None
storage_path	`string`	Local only — directory where indexed documents are stored.	`./.pageindex`
index_backend	`dict`	Local only — connection overrides for the indexing lane’s LLM calls (LiteLLM connection params).	None
chat_backend	`dict`	Default connection overrides for the chat surfaces. A call’s own `backend` keys win over it.	None
summary_model	`string`	Local only — legacy override for node summaries and document descriptions; `index_model` covers this.	None
retrieve_model	`string`	Legacy name for `chat_model`.	None

Model naming

Model names follow LiteLLM’s naming convention . Choose the format that matches your provider.

OpenAI — use the model name directly and set OPENAI_API_KEY:


os.environ["OPENAI_API_KEY"] = "your-openai-api-key"
chat_model = "gpt-5.6-sol"

Anthropic — prefix with anthropic/ and set ANTHROPIC_API_KEY:


os.environ["ANTHROPIC_API_KEY"] = "your-anthropic-api-key"
chat_model = "anthropic/claude-sonnet-4-6"

OpenRouter — prefix with openrouter/ and set OPENROUTER_API_KEY:


os.environ["OPENROUTER_API_KEY"] = "your-openrouter-api-key"
chat_model = "openrouter/anthropic/claude-sonnet-4-6"

Bare names are OpenAI-compatible shorthand, so any OpenAI-compatible server works through the usual OPENAI_API_KEY / OPENAI_BASE_URL configuration. Write openai/Qwen/... when the server itself serves slashed model ids (vLLM, TGI).

For other providers, see the LiteLLM provider documentation .

Backend overrides

index_backend and chat_backend carry connection details — keys, base URLs, cloud credentials — without touching environment variables. Keys are passed through verbatim to the underlying client.


client = PageIndexClient(
    index_model="gpt-5.6-luna",
    chat_model="gpt-5.6-sol",
    chat_backend={"api_key": "sk-...", "base_url": "https://my-gateway.internal/v1"},
)
 
# Per-call overrides win over the client default
client.chat_completions(messages, doc_id=doc_id, backend={"api_key": "sk-other"})

The vocabulary follows whichever door runs: LiteLLM connection params for chat_completions(), the OpenAI SDK’s client params for responses(), the Anthropic SDK’s for messages(). api_key and base_url mean the same thing on all three.

Cloud-only surfaces

These raise PageIndexAPIError in local mode:

Folders — create_folder(), list_folders(), and folder_id filters
beta_headers on submit_document()
The deprecated retrieval API — submit_query(), get_retrieval()
enable_citations=True on chat_completions() (managed chat only)

🔧 PageIndexClient

Constructor parameters

Model naming

Backend overrides

Cloud-only surfaces

💬 Community & Support