Interactive matcher

Moderation Stack Matcher

Rank moderation tools and stacks against your modalities, harm classes, volume, latency budget and deployment limits, with dated and sourced cost notes.

Hard constraints (modality, deployment, latency, sunset status) filter the candidate set; remaining tools are ranked by a weighted harm-class coverage score against your budget. Everything runs in your browser and your spec is encoded in the URL. Coverage and cost figures are read from vendor documentation and published evaluations, and are summarised in our tool reviews (last reviewed 2026-09).

What content types do you moderate?

A tool must cover every modality you check, or it is filtered out.

Which harm classes must you cover?

Weighted into the coverage score; tools with a hard zero on a required class are penalized heavily.

Volume tier (items per month)

Scales the cost estimate; very high volume favors cheap self-hosted options.

Thousands / mo Millions / mo Billions / mo

Latency budget

Sync inline moderation needs sub-300ms tools; async tolerates heavier models.

Synchronous — must block inline, < 300ms Asynchronous — review out of band is fine

Deployment constraint

A hard filter — air-gapped excludes every hosted API.

API is fine — calling a vendor service is acceptable Must self-host — no traffic to third parties Air-gapped — fully offline, no external dependencies

Budget sensitivity

Weights cost-per-million in the final ranking.

Cost-sensitive — minimize spend Balanced — cost matters but coverage wins Coverage-first — budget is not the constraint

All 12 tools & stacks we score

Coverage is graded per harm class: 2 = strong, 1 = partial, 0 = none. Cost-per-million is a rough order-of-magnitude estimate (USD), not a quote.

Tool / stack	Deploy	Latency	~$ / M
Llama Guard 3 (8B, self-hosted) ↗	selfhost	async	$6
Llama Guard 3 Vision (11B) ↗	selfhost	async	$14
ShieldGemma 1 (9B text, self-hosted) ↗	selfhost	async	$6.5
OpenAI Moderation API (omni-moderation) ↗	api	sync	free
Perspective API (sunsetting, ends 2026-12-31) ↗	api	sync	free
AWS (Comprehend toxicity + Rekognition) ↗	api	sync	$1000
Azure AI Content Safety ↗	api	sync	$375
NeMo Guardrails (self-hosted framework) ↗	selfhost	async	$8
Microsoft Presidio (PII) ↗	selfhost	sync	$2
Stack: OpenAI Moderation + Presidio ↗	api	sync	$2
Stack: Llama Guard + Prompt Guard + Presidio (self-hosted) ↗	selfhost	async	$9
Stack: Azure Content Safety + classifier ensemble ↗	api	sync	$385

Related tools in this network

Other interactive tools across the network that pair well with this one.