---
title: "Get Llm Proxy Requests For Call"
method: GET
path: "/developer/llm-proxy-requests/call/{vapi_call_id}"
tags: ["Developer", "Developer - LLM Proxy"]
---

# Get Llm Proxy Requests For Call

`GET /developer/llm-proxy-requests/call/{vapi_call_id}`

## Request

- Base URL: `https://api.openintake.ai` (the document declares no server; this is the origin it was published from)
- URL: `https://api.openintake.ai/developer/llm-proxy-requests/call/{vapi_call_id}`
- Auth: none declared

One call's requests in order, with the computed ``request_seq``.

``request_seq`` orders reading; it is not a join key against
``vapi_turn_latency.turn_index``. Vapi retries a failed turn 3x, a tool-calling turn
is one Vapi turn but 2+ requests, and a model-generated first message is a request
with no user turn at all. Align the two positionally with a FULL OUTER JOIN so
divergence shows up as one-sided NULLs instead of being silently dropped.

## Path parameters

- `vapi_call_id` string, required

## Response `200`

Successful Response

- LLMProxyRequestListResponse
  - `items` LLMProxyRequestItem[]
    - `request_id` string, required
    - `vapi_call_id` string, nullable
    - `call_id` string, uuid, nullable
    - `customer_id` string, uuid, nullable
    - `logged_at` string, date-time, required
    - `outcome` string, nullable
    - `winner` string, nullable
    - `winner_ttft_ms` number, nullable
    - `total_ms` number, nullable
    - `stream_duration_ms` number, nullable
    - `assistant_hint` string, nullable
    - `requested_model` string, nullable
    - `effective_model` string, nullable
    - `model_choice_reason` string, nullable
    - `n_messages` integer, nullable
    - `n_tools` integer, nullable
    - `output_chunks` integer, nullable
    - `output_chars` integer, nullable
    - `finish_reason` string, nullable
    - `stop_reason` string, nullable
    - `tool_names` string[], nullable
    - `error` string, nullable
    - `anthropic_ttft_ms` number, nullable
    - `bedrock_ttft_ms` number, nullable
    - `anthropic_error_kind` string, nullable
    - `bedrock_error_kind` string, nullable
    - `image_tag` string, nullable
    - `ecs_task_id` string, nullable
    - `prewarm_ok` boolean, nullable
    - `has_bodies` boolean: Whether a bodies line was imported for this request
    - `request_seq` integer, nullable
  - `total` integer: Total matching rows, ignoring pagination

## Other responses

- `422`: Validation Error

## Changes

- **2026-09-20** `28867c1dbb52`: 2 info
  - added the optional property `items/items/effective_model` to the response with the `200` status
  - added the optional property `items/items/model_choice_reason` to the response with the `200` status

[Operation history](https://skmtc.dev/openintake/apis/openintake-api/changes/developer/llm-proxy-requests/call/:vapi_call_id/get.md)

---

[API](https://skmtc.dev/openintake/apis/openintake-api.md) - [All operations](https://skmtc.dev/openintake/apis/openintake-api/llms.txt) - [OpenAPI document](https://skmtc.dev/openintake/apis/openintake-api/revisions/de90a8dcbc07?raw)