> ## Documentation Index
> Fetch the complete documentation index at: https://api.thesapientcompany.com/llms.txt
> Use this file to discover all available pages before exploring further.

# What's new

> Everything recently shipped in the Sapient API — pricing, lenses, the wallet endpoint, and the expanded manipulation read.

A running summary of what's currently live. Everything below is shipped and reflected in the rest of the reference.

## Pricing

* **\$2.50 per scan** on [Qualia](/api-reference/models) — the default and only available model right now. Charged **once** when a scan completes; failed scans are never charged. (Mary is temporarily unavailable.)
* The **first \$2.50 of usage each month is free** — exactly one Qualia scan — then scans debit your prepaid deposit. See [Pricing](/api-reference/pricing).
* Consumer (dashboard) plans are **metered**: Free **1**, Pro **60**, and Studio **150** scans. The API itself is usage-based and requires a paid plan to mint a key.

## Lenses

* **4 selectable lenses**: `attention`, `memory`, `purchase_intent` (a.k.a. Buy / Not Buy), and `manipulation`. There is no `emotion` lens. See [Lenses](/api-reference/lenses).

## New: wallet endpoint

* **`GET /v1/wallet`** returns your spendable balance — deposit, this month's free credit, the per-scan price, and an estimate of how many scans you can still run — so an agent or script can check funds before submitting. Read-only; it never debits. See [Check your wallet](/api-reference/wallet).

## Expanded manipulation outputs

When a scan is scored under the `manipulation` lens, the `manipulation` block (on both [`GET /v1/scans/{id}`](/api-reference/get-scan) and `POST /v1/analyze`) now adds:

* **`pbi`** — the Prefrontal Balance Index (vmPFC − dlPFC): how reward/emotion-driven the response is relative to deliberate reasoning.
* **`threat`** — the headline threat verdict shown in the UI (`label`, `severity`, `color`, `note`).
* **`population`** — a **directional** population-impact projection (`per_ten`, `per_ten_thousand`) — a model estimate, not a measured outcome.
* **`timeline`** — the per-second guard-down timeline (`exposure = reward − effort − alarm`).

These join the existing `conversion_risk`, `risk_band`, `cognitive_sovereignty_index`, `sovereignty`, `quadrant`, `circuits`, `mechanisms`, `peak_moment`, and `tactics`. See the full shape in [Get a scan](/api-reference/get-scan#param-manipulation).

## Other

* **3-minute video cap.** Clips longer than 3 minutes are rejected up front (no GPU spend, no charge). Trim or split before submitting.
* **One-line MCP / CLI install.** Wire Sapient into Claude Code, Cursor, or any MCP agent with a single command, or run scans from the terminal with the CLI. See [Connect an agent](/connect-an-agent).