Jev versus moderation APIs: a practical comparison
A moderation API returns a vendor’s fixed categories (hate, sexual, self-harm, …). Jev returns answers to your questions: “violates our marketplace weapons policy,” “looks like a chargeback threat.”
Unofficial. We do not smear hosted moderators and we do not invent category AUCs. docs.typesafe.ai. No keys sold here (and not theirs either).
Comparison scope
If you need a standard safety taxonomy with a vendor’s published eval story, start there. If you need a house policy that is not in their enum, add Jev — or a trained classifier.
Criteria that decide the architecture
| Axis | Jev (System One) | Moderation APIs |
|---|---|---|
| Label set | Your criteria (Choice ≤ 255 options) |
Vendor enum |
| Versioning | Pin jev-1.13.0 + git the questions |
Vendor model revisions |
| On-prem | Only if your channel allows; usually API | Sometimes |
| Price | Jev input tokens (vendor $0.042/M) | That API’s card |
Decision quality and control
Fixed taxonomies are easier to compare across apps. Custom criteria are easier to align with counsel’s memo. Neither is “can’t be wrong.”
Integration trade-offs
Run the hosted moderator for the categories they attest; run Jev for house rules; AND the blocks. Do not average unrelated scores.
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
- Your policy is not in the vendor enum
Prefer Moderation APIs when
- You need their attested categories or a procurement checkbox
Vendor taxonomy ∩ house Jev questions. Code ANDs denials.
What this page does not claim
-
No head-to-head toxicity F1.
-
Schema-safe output is not the same as factual correctness.
-
No independent bake-off numbers live here.
FAQ
Can I fine-tune Jev on our moderation labels? TypeSafe: Jev is not LoRA-adapted per customer. Shape answers via state and criteria.
Noul vs Choice for “violates policy”? Noul = P(yes). Choice if you need a category. Do not reuse a Noul threshold on a Choice.
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 content filters, usecase guardrails policy checks, glossary guardrail. Canonical: https://docs.typesafe.ai.
Sources
Public TypeSafe or adjacent documentation only. No private claims.