---
title: "List Models"
method: GET
path: "/api/v1/vision/models"
tags: ["Vision"]
---

# List Models

`GET /api/v1/vision/models`

List every model Argus supports, with pricing.

Two families: the curated VLMs served by /vision/locate (per-token
pricing; pass their id as `model`) and the Axilio model line behind
/vision/detect (per-page pricing; selected via `ocr_engine` /
`inference_type` — lite is the free engine, pro is premium).

Public: no API key required (it's a catalog of model names and public
prices, nothing sensitive), so a client can discover supported models
before it holds credentials. The SDK fetches this once, caches it, and
validates find(model=...) locally so a typo fails fast with a clean
error instead of a 400 from /locate.

## Response `200`

Successful Response

- SupportedModelsResponse — Response for GET /vision/models. Lists every model Argus supports, each with its final post-markup pricing: the curated VLMs for the element-locator task (per-token pricing; pass their id as `model` to /vision/locate) and the Axilio model line behind /vision/detect (per-page pricing; selected via the `ocr_engine` and `inference_type` request fields — lite = the free engine, pro = premium). Lets a client compare cost and validate `find(model=...)` up front instead of round-tripping to /locate and getting a 400. The VLM set is curated: `/locate` enforces it — an unsupported model is rejected, so this is the authoritative allowlist, not just advisory. The Axilio ids are not /locate targets. Shaped like a standard /v1/models response ({object: "list", data: [...]}).
  - `object` 'list'
  - `data` ModelInfo[], required — Supported models, sorted by id.
    - `id` string, required — Model identifier, e.g. 'anthropic/claude-sonnet-4.5' or 'axilio/argus-ocr-pro-1'. VLM ids are passed as `model` to /vision/locate; Axilio ids identify the engine behind /vision/detect calls (selected via `ocr_engine` / `inference_type`) and are what usage rows are labeled with.
    - `object` 'model'
    - `name` string, required — Human-readable model name.
    - `owned_by` string, required — Model provider, e.g. 'anthropic' or 'axilio'.
    - `type` string, required — Model type: 'vlm' for the vision-language models served by /vision/locate; 'detect', 'ocr', and 'vision' for the Axilio model line behind /vision/detect.
    - `context_window` integer, required — Maximum context window, in tokens.
    - `pricing` ModelPricing, required — The customer's billed rates in USD. VLMs price per token (input/output); the Axilio model line prices per page, where one screenshot is one page — so per_page reads as the price of one call. A rate is "0" when that unit is free for the model.
      - `input` string, required — USD per input token.
      - `output` string, required — USD per output token.
      - `per_page` string, nullable — USD per page (one screenshot = one page) for the per-page-billed Axilio models. Absent for per-token models.

---

[API](https://skmtc.dev/axilioai/apis/argus-2.md) · [All operations](https://skmtc.dev/axilioai/apis/argus-2/llms.txt) · [OpenAPI document](https://skmtc-service-production.skmtc.workers.dev/v1/apis/axilioai/argus-2/revisions/6b923d4d3201/schema)