# SightRadar vs Face++

> Face++ (by Megvii) is a long-running face platform with a broad feature catalog, but its commercial model leans on free-tier QPS limits and paid concurrency packages you size up front. SightRadar focuses on the core recognition workflow — index, search, compare, detect — on an AWS Rekognition–compatible API, with flat prepaid per-photo pricing ($0.00093/photo, no QPS tiers to reason about), a real web console, ready-made SDKs, and per-tenant data isolation.

Source: https://sightradar.com/compare/face-plus-plus

## Bottom line

For core face recognition, SightRadar wins on price (flat prepaid per-photo, no QPS packages), reliability (no concurrency tier to outgrow), and the web UI (a real collection browser with per-operation cost). Pick Face++ only if you specifically need its wide catalog of extra facial-analysis features.

## Headline verdicts

### Price — winner: SightRadar
- SightRadar: Flat $0.00093/photo prepaid — no QPS tiers, no concurrency package to pre-buy.
- Face++: Free tier with QPS limits; paid throughput depends on concurrency packages you size in advance.

### Reliability & access — winner: SightRadar
- SightRadar: Predictable throughput with no QPS package to reason about; per-tenant isolation by design.
- Face++: Throughput is bounded by the concurrency package; free-tier QPS is tight.

### Web UI — winner: SightRadar
- SightRadar: Console to browse collections, run searches, manage keys, and see per-operation usage and cost.
- Face++: Developer console exists, but the model centers on API keys, QPS, and packages rather than a collection browser with cost breakdown.

## Where SightRadar wins

- Flat, predictable pricing: $0.00093/photo prepaid with no QPS tiers and no concurrency package to pre-buy and outgrow.
- AWS Rekognition–compatible API plus ready-made zero-dependency Python and Node SDKs — migrate without rewriting calls.
- A real web console: browse collections, run a search in the playground, and see per-operation usage and exact credit cost.
- Per-tenant isolation by design, with privacy terms written specifically for a face-recognition processor and clear data residency.

## Where Face++ is the better fit

- Face++ offers a much broader catalog of facial-analysis features (dense landmarks, attributes, liveness, body/gesture) — if you need far more than core recognition, that breadth is its real advantage.

## Feature-by-feature

### Getting started

| Feature | SightRadar | Face++ |
| --- | --- | --- |
| Rekognition-compatible API | ✅ Drop-in for AWS Rekognition face calls | ❌ Its own API surface |
| Ready-made SDKs | ✅ First-party zero-dependency Python & Node/TS SDKs | ➖ REST API with community/SDK wrappers |

### Pricing

| Feature | SightRadar | Face++ |
| --- | --- | --- |
| Pricing model | ✅ Flat prepaid $0.00093/photo, no QPS tiering | ➖ Free tier with QPS limits; paid concurrency packages |

### Reliability & access

| Feature | SightRadar | Face++ |
| --- | --- | --- |
| Predictable per-call throughput | ✅ No concurrency package to size up front | ➖ Throughput depends on the purchased concurrency package |

### Web UI & visibility

| Feature | SightRadar | Face++ |
| --- | --- | --- |
| Collection browser + cost view | ✅ Browse collections and see per-operation credit cost | ➖ Console centers on keys/QPS/packages, not a cost breakdown |

### Core capability

| Feature | SightRadar | Face++ |
| --- | --- | --- |
| Per-tenant isolation | ✅ Vectors never cross-searchable between accounts | ➖ FaceSets scoped to your API key |
| Calibrated similarity scores | ✅ Threshold on calibrated scores | ✅ Confidence with provider thresholds |

Competitor pricing changes over time — verify the current rate at https://www.faceplusplus.com/pricing-details/. SightRadar's rates are at https://sightradar.com/pricing.

## Try SightRadar

Self-serve sign-up, prepaid credits, AWS Rekognition–compatible API, ready-made Python & Node SDKs, and a real web console. Start free at https://sightradar.com/login — docs at https://sightradar.com/docs.