How many answers were correct
Each bar is one benchmark: how many items it answered correctly out of every item in that set. We wrote the questions and answer keys ourselves and had them checked by two native speakers, and every model answers exactly the same items.
8 benchmarks
Pemahaman slang Indonesia37/42 88.1%
Normalisasi teks Indonesiaquestion version 3KeteranganWe asked this benchmark with several different phrasings of the question, on the same items and the same answer key. Scores across phrasings are not directly comparable; open the benchmark page to see them side by side.30/49 61.2%
Koreksi ejaan Indonesia23/38 60.5%
Tingkat tutur bahasa Jawa16/27 59.3%
Normalisasi teks Indonesiaquestion version 2KeteranganWe asked this benchmark with several different phrasings of the question, on the same items and the same answer key. Scores across phrasings are not directly comparable; open the benchmark page to see them side by side.28/49 57.1%
Normalisasi teks Indonesiaquestion version 1KeteranganWe asked this benchmark with several different phrasings of the question, on the same items and the same answer key. Scores across phrasings are not directly comparable; open the benchmark page to see them side by side.19/49 38.8%
Arti kata Jawa sehari-hari52/135 38.5%
Arti kata Jawa langka4/43 9.3%
How far the cost pays off
Sorted by the cheapest correct answer first. The left column shows how many items were answered correctly, so a top row with a long left bar means both cheap and accurate.
8 with cost on record
Correct answersCost per correct answer
Normalisasi teks Indonesiaquestion version 3KeteranganWe asked this benchmark with several different phrasings of the question, on the same items and the same answer key. Scores across phrasings are not directly comparable; open the benchmark page to see them side by side.61.2%$0.000012
Normalisasi teks Indonesiaquestion version 2KeteranganWe asked this benchmark with several different phrasings of the question, on the same items and the same answer key. Scores across phrasings are not directly comparable; open the benchmark page to see them side by side.57.1%$0.000013
Normalisasi teks Indonesiaquestion version 1KeteranganWe asked this benchmark with several different phrasings of the question, on the same items and the same answer key. Scores across phrasings are not directly comparable; open the benchmark page to see them side by side.38.8%$0.000017
Tokens used
Input and output are kept apart because they are priced differently, often tenfold. A long output bar means the model talks a lot, and that is where nearly all the cost difference comes from.
94,454 in, 8,599 out
inputoutput
Pemahaman slang Indonesia13,796 · 125
Normalisasi teks Indonesiaquestion version 3KeteranganWe asked this benchmark with several different phrasings of the question, on the same items and the same answer key. Scores across phrasings are not directly comparable; open the benchmark page to see them side by side.13,812 · 860
Koreksi ejaan Indonesia13,819 · 226
Tingkat tutur bahasa Jawa6,781 · 78
Normalisasi teks Indonesiaquestion version 2KeteranganWe asked this benchmark with several different phrasings of the question, on the same items and the same answer key. Scores across phrasings are not directly comparable; open the benchmark page to see them side by side.14,502 · 849
Normalisasi teks Indonesiaquestion version 1KeteranganWe asked this benchmark with several different phrasings of the question, on the same items and the same answer key. Scores across phrasings are not directly comparable; open the benchmark page to see them side by side.12,501 · 853
Arti kata Jawa sehari-hari15,811 · 4,563
Arti kata Jawa langka3,432 · 1,045
Wait time per call
Median wait for a single call on each benchmark. Median rather than mean, so one stalled call does not move the number.
8 with wait time on record
Normalisasi teks Indonesiaquestion version 3KeteranganWe asked this benchmark with several different phrasings of the question, on the same items and the same answer key. Scores across phrasings are not directly comparable; open the benchmark page to see them side by side.2.27 s
Normalisasi teks Indonesiaquestion version 1KeteranganWe asked this benchmark with several different phrasings of the question, on the same items and the same answer key. Scores across phrasings are not directly comparable; open the benchmark page to see them side by side.2.68 s
Normalisasi teks Indonesiaquestion version 2KeteranganWe asked this benchmark with several different phrasings of the question, on the same items and the same answer key. Scores across phrasings are not directly comparable; open the benchmark page to see them side by side.5.34 s
Arti kata Jawa langka6.81 s
Koreksi ejaan Indonesia33.81 s
Test coverage
6/ 6
This model has been tested on all 6 benchmarks we publish.
Earlier question versions
The questions or the key changed after these were measured. Kept as history, not as current results.show 2 rowsclose
Pemahaman slang Indonesia24/27 88.9%
Arti kata Jawa sehari-hari52/135 38.5%
Log10 calls to this model, spread across 3 days.OpenClose
This model is called through a provider id that may not pin a version, so the dates below state which weights were actually tested. Rows marked with a hyphen have a date we inferred from file history rather than recorded at call time, so we do not show a clock time for them.
1 runs
- 22:19Pemahaman slang Indonesiamultiple choice37/42
1 runs
- 20:06Arti kata Jawa sehari-hariword meaning52/135
8 runs
- 11:00Arti kata Jawa sehari-hariword meaning52/135
- 10:38Normalisasi teks Indonesiaopen answerPrompt stating the key's rule30/49
- 10:20Normalisasi teks Indonesiaopen answerDetailed prompt forbidding word swaps28/49
- 09:59Normalisasi teks Indonesiaopen answerBuilt-in prompt19/49
- 09:39Arti kata Jawa langkaword meaning4/43
- 09:26Tingkat tutur bahasa Jawamultiple choice16/27
- 09:14Koreksi ejaan Indonesiaopen answer23/38
- 08:36Pemahaman slang Indonesiamultiple choice24/27