Jev versus semantic search: a practical comparison
Semantic search returns nearby passages. Jev can score or choose among those passages. It is not an index and it is not a query planner.
Unofficial. TypeSafe’s line-by-line and rerank cookbooks are the official worked examples — lifts they publish stay on those pages (vendor-measured). docs.typesafe.ai.
Comparison scope
If you need “find the paragraph,” start with BM25/dense search. If you need “is this the answer, or none,” add a Choice + Noul as in their semantic-find cookbook.
Criteria that decide the architecture
| Axis | Jev (System One) | Semantic search |
|---|---|---|
| Corpus size | Bounded by context (64k request / 32k state+longest — vendor) | Millions of docs |
| Recall job | Poor substitute for retrieval | Wins |
| Precision on a shortlist | Score/Noul per pair | Optional reranker |
| “None of the above” | You can add it | Often implicit |
Decision quality and control
Dumping a wiki into state hits jaggedness (irrelevant detail) and the token bill. Retrieve first. Their rerank cookbook shortlists with BM25 then one question per pair — that layering is the point.
Integration trade-offs
query → retrieve K → Jev per candidate or one big Choice with none → return ids. Generation of the answer is still an LLM if you need prose.
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
- You already have a shortlist and need a keep/drop/flag
Prefer Semantic search when
- You must search a large corpus
Search then decide. See also vs RAG.
What this page does not claim
-
We do not restated cookbook percentages as ours.
-
Schema-safe output is not the same as factual correctness.
-
No independent bake-off numbers live here.
FAQ
Is Jev a search engine? No.
218-way Choice? TypeSafe shows a large Choice plus a Noul “does an answer exist?”. Still not an index.
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 vector databases, vs retrieval only search, rag passage classification. Canonical: https://docs.typesafe.ai.
Sources
Public TypeSafe or adjacent documentation only. No private claims.