Perspective API
JigsawJigsaw's free hosted API for scoring the toxicity of text and related attributes such as insult, threat and identity attack. It has been used for comment moderation since 2017. Jigsaw has announced that the service is sunsetting and will be out of operation after 2026.
The service runs until 31 December 2026; new usage and quota requests closed in February 2026 and no migration support is offered. The reason given is that AI capabilities have evolved and demand for a standalone tool in this area has fallen. Verified 2026-08-13 via the Perspective API site.
Openness
1 high confidence- service
- proprietary hosted API by Jigsaw/Google(no weights, no self-host)
- access
- free with quota
- scores
- toxicity + related attributes
- source
- closed
- license
- proprietary
Closed, API-only hosted classifier; no weights or source. Long-standing and influential, but proprietary. The product is also ending - the site leads with a sunset notice and the API goes out of service after 2026 - but that is a fact about availability rather than about openness, so the closed score stands.
- https://perspectiveapi.com/ recorded 2026-08-13
Jigsaw's hosted API, governed by its own API Terms with no source or self-hostable implementation offered anywhere on the site; access is by quota, and the page handles usage and increased-quota requests. It now opens 'Important Notice - Perspective API is sunsetting and service is officially ending after 2026', with 'The service will remain active until December 31, 2026' and no migration support.
Adoption
4 medium confidenceJigsaw has never published a download, install or request count for Perspective, so the level rests on reported standing rather than on a measured figure, and no numeric reach is claimed: a 1M-10M label would assert a count nobody made. The site also announces that the API is sunsetting and goes out of service after December 31, 2026, giving as the reason that 'AI capabilities have evolved, and there is now less demand for a standalone tool specific to this area'. Level 4 describes the standing the cited sources show and none of them argues for a different one - though a level 4 on a free API that is leaving service describes the past more than the present.
- https://perspectiveapi.com/ recorded 2026-08-13
Sunset announcement and transition period. No usage figure of any kind - no request volume, no customer count, no developer count - appears on the page.
Capability
3 low confidenceMature toxicity scoring with broad real-world use; narrower than full safety taxonomies and policy-conditioned models. The main product page no longer carries the attribute list, having been replaced by the sunset notice, and the developer site that does carry it is a client-rendered Salesforce Experience page serving an empty body that no change of headers fixes - so the facts come from elsewhere. The Comment Analyzer discovery document, which is the API's own machine-readable definition and is still served live, shows the service running and multilingual, taking "ISO and BCP-47 language codes" and detecting language per comment. The fixed attribute set is corroborated by Lasso's write-up, which names it exactly - "The production attributes (TOXICITY, SEVERE_TOXICITY, IDENTITY_ATTACK, INSULT, PROFANITY, THREAT)" - plus experimental ones. Lasso sells a competing moderation product, so that is a document rather than a re-derivable dataset, and confidence stays low. The same page also states that Perspective "will sunset after December 31, 2026, with no migration path provided", which sits awkwardly beside a live capability band.
- https://perspectiveapi.com/ recorded 2026-08-13
Sunset notice only. The toxicity and attribute scoring the value describes is no longer documented anywhere on this page.
- https://commentanalyzer.googleapis.com/$discovery/rest?version=v1alpha1 recorded 2026-08-14
the API's own discovery document, live at 200 - a comment-scoring service whose requests carry "language(s) of the comment and context", accepting "Both ISO and BCP-47 language codes", with per-comment language detection
- https://www.lassomoderation.com/blog/perspective-api-toxicity/ recorded 2026-08-14
"The production attributes (TOXICITY, SEVERE_TOXICITY, IDENTITY_ATTACK, INSULT, PROFANITY, THREAT) are stable and well-tested", with experimental attributes beyond them; and "Google announced that Perspective API will sunset after December 31, 2026"
Verified 2026-08-13