Wire contract, primitive boundaries, and operations

E01–E16 · DESIGN FILE, AS WRITTEN

16 experiments, 670 records and 642,098 input tokens.

C

Confirmed

12

F

Falsified

3

M

Mixed

1

We sent Jev deliberately broken, oversized, and edge-case requests to find out where it actually breaks and what it tells you when it does.

Reading the record…

Experiments in this group

Exp

Verdict

Question

E01

C

Confirmed

How many levels can a Score question have before Jev refuses it?

E02

C

Confirmed

How many options can a Choice question have before Jev refuses it?

E03

C

Confirmed

Does changing the order of a question's options change the answer?

E04

C

Confirmed

Can other questions in the same request change an answer?

E05

F

Falsified

Does Jev's confidence number always run below the top probability?

E06

C

Confirmed

Can you calculate Jev's confidence number yourself from the probabilities?

E07

M

Mixed

Does Jev's response say exactly which model version answered?

E08

C

Confirmed

What error codes does Jev actually return for a bad request?

E09

C

Confirmed

Do you have to include instructions text on every question?

E10

C

Confirmed

What in a request actually drives its token cost?

E11

C

Confirmed

What happens if a request goes over the token budget?

E12

F

Falsified

Does asking Jev the exact same question twice give the exact same answer?

E13

F

Falsified

Do Jev's answers always follow their own basic math rules?

E14

C

Confirmed

Does renaming a question or its answer keys change the result?

E15

C

Confirmed

How many requests at once before Jev starts rejecting them?

E16

C

Confirmed

Does asking more questions in one request save money without slowing it down?

108 experiments, 38,194 API responses logged, 45,861,280 input tokens, $1.9262 at $0.042 per million.

Compiled from dependency-docs-reference/jev/notes/12-experiments on 2026-09-19.