The oddness wrapper has a smaller JavaScript entry and request body. Its complete archive is larger, and its live benchmark did not establish lower latency.
is-jodd asks Jev whether an integer is odd. is-jeven asks whether it is even. For callers comparing oddness, the relevant expressions are await isOdd(n) and !(await isEven(n)). Both depend on a remote model; neither should replace exact local arithmetic when correctness is the goal.
This is a comparison from is-jodd’s publisher, not an independent laboratory or official TypeSafe statement. The figures below describe version 1.0.0 of both packages and our recorded experiment.
The shared public contract
Both wrappers accept a safe integer, use an API key, support an abort signal, and resolve to a boolean. Both have zero runtime dependencies. The important implementation difference is the question: is-jeven uses Choice with evenness options, while is-jodd uses Noul with a strict probability cutoff above 0.5 for oddness.
The negation must happen after awaiting the evenness result:
const oddFromEven = !(await isEven(37, options));
const oddDirectly = await isOdd(37, options);
This illustrates equivalent intentions, not a guarantee that two model requests agree. See await before negating for the Promise trap.
Smaller depends on what you measure
is-jodd’s published JavaScript entry is 861 bytes versus 1,437 bytes, a 40.1% reduction. Gzipping those entries at level nine yields 562 versus 748 bytes. Our release build minifies the entry, which contributes to the difference.
The entire npm archive tells a different story: 2,870 bytes for is-jodd versus 1,803 bytes for is-jeven. is-jodd’s archive includes its custom license and other documentation and metadata. Calling the whole package “40% smaller” would misdescribe the measurement.
For input 3, the serialized request bodies are 108 and 159 bytes. The 51-byte saving was consistent across our tested inputs; the percentage changes with integer length. These counts exclude headers and TLS.
The live result was inconclusive on speed
We measured twenty matched input pairs after two warmup pairs, making 44 requests. Mean times were 157.51 milliseconds for is-jodd and 155.62 milliseconds for the awaited is-jeven negation. is-jodd won nine pairs. Both answered all twenty measured inputs correctly, with no recorded errors or timeouts.
The exploratory interval for mean paired savings crossed zero. That does not support a general faster claim, and twenty correct answers do not establish guaranteed accuracy. Read the benchmark methodology for the environment and raw-record links.
Keep the practical differences visible
is-jodd has a restrictive JODD-1T license requiring $1 trillion and written permission before use unless separate terms are granted. It is not MIT. For ordinary work, the calculator or a validated remainder check gives the answer without these package and service considerations.