Jev versus content filters: a practical comparison
Content filters are deny-lists, regex, and classifier scores with frozen labels. Jev lets you describe this product’s hazards in criteria without training a new filter model.
Independent. Official guardrails cookbook: screen inbound/outbound with Nouls and Scores, then you threshold. Not a safety-product claim. docs.typesafe.ai.
Comparison scope
CSAM, malware, and legal-must-block classes belong on purpose-built filters and vendor stacks with attested evals. Jev is an extra semantic layer for policy you can write in English.
Criteria that decide the architecture
| Axis | Jev (System One) | Content filters |
|---|---|---|
| Exact token ban | Overkill / weaker | Wins |
| Novel phrasing of a policy | Noul per hazard | Misses |
| Attested safety evals | None from jev.pro | Specialist vendors |
| Adversarial input | Jaggedness: not hostile-by-default | Varies |
Decision quality and control
A filter that must never fail closed-loop on illegal content should not have Jev as the only control. Use Jev to rank residual risk after the hard filter.
Integration trade-offs
Hard filter → Jev hazard Nouls + severity Score → code: pass / review / block. Log all three layers.
TypeSafe’s public models page lists jev-1.13 at $0.042 per million input tokens with output tokens free — a vendor claim, not a jev.pro measurement. Confirm on the models page before you bid.
When each approach fits
Prefer Jev when
- Policy is specific and changes with the product
Prefer Content filters when
- You need an exact string ban or a certified classifier
Stack, don’t swap. See moderation APIs.
What this page does not claim
-
Not a complete Trust & Safety program.
-
No claimed block rates.
-
Schema-safe output is not the same as factual correctness.
-
No independent bake-off numbers live here.
FAQ
Is Jev a moderation API? It can score custom questions. It is not a drop-in Perspective/OpenAI Moderation clone.
Jailbreaks? TypeSafe publishes a prompt-injection-oriented cookbook. Test. Not a guarantee.
Disclaimer
This is an independent unofficial site and is not affiliated with TypeSafe AI; official documentation is available at https://docs.typesafe.ai. Never treat jev.pro as TypeSafe official documentation. We do not sell, issue, or proxy API keys.
Hub: Comparisons. Siblings: vs moderation apis, vs keyword triage, llm guardrails. Canonical: https://docs.typesafe.ai.
Sources
Public TypeSafe or adjacent documentation only. No private claims.