Decision model
Jev 1.14.0
1 question per request
59.2% of the 2,198 questions right (57.1% to 61.2%, 95% interval). Back to the leaderboard.
How we called it
- Kind
- Decision model
- Access
- TypeSafe System One API
- Paid as
- Provider service
- Model ID
jev-1.14.0- Confidence from
- Option score
- Run
- 194, 16 October 2026
Settings
- Reasoning
- off
- Request format
- 1 question per request
- Requests at once
- 1
- Temperature
- not recorded
- Harness seed
- 0
- Harness commit
5056b43e3a36- Prompt version
254597b8be16
Results
| Questions | Accuracy | 95% interval | Right |
|---|---|---|---|
| All questions | 59.2% | 57.1% to 61.2% | 1,301 of 2,198 |
| Reading | 85.1% | 82.1% to 87.6% | 558 of 656 |
| Grammar | 48.2% | 45.7% to 50.7% | 743 of 1,542 |
Comparison group: wxyzA. Compare with another setup.
Accuracy by topic
Page images (preliminary)
Preliminary: these numbers are not final and may change. This setup is condition A of the Page images tab, which asks the 763 questions with underlined or numbered words as text (A), as a page image (B), as both (C) and through OCR models (D); the share answered right, with its 95% interval.
| Model | A · text onlyaccuracy, higher is better | B · image onlyaccuracy, higher is better | C · text plus imageaccuracy, higher is better | Best OCR routeaccuracy, higher is better |
|---|---|---|---|---|
| Jev 1.14.01 question per request | 54.3%50.7% to 57.8% | 43.3%39.8% to 46.8% | 58.6%55.1% to 62.0% | 46.4%via Surya OCR 242.9% to 49.9% |
Confidence
- Answers kept at 99% or more
- 260, of which 1 wrong (share right: 97.9% to 99.9%, 95% interval); 1,938 skipped, the model less than 99% sure
- Calibration gap
- 2.5 points
- AUROC, all questions
- 0.79 0.77 to 0.81
- AUROC, reading
- 0.88 0.85 to 0.91
- AUROC, grammar
- 0.70 0.67 to 0.72
- Brier score
- 0.180
- Accuracy of its surest answers
- most confident fifth: 98.6% right; most confident half: 81.3% right; most confident four fifths: 66.0% right
- Typical score of a right answer
- 76.0%
- Typical score of a wrong answer
- 41.0%
Repeats
Run 5 times on all questions.
- Answers changed in 5 runs
- 291 of 2,198
- Same wrong answer in all 5
- 719 of 2,198
- Wrong in at least one run
- 1,010 of 2,198
- Accuracy in each run
- 59.2%, 58.2%, 58.8%, 57.8%, 58.3%
Latency
- Latency, median request
- 0.211 s (this run)
- Latency, 90th percentile
- 0.277 s
- Requests at once
- 1
- Throughput
- 265.7 questions per minute of request time
- Request time
- 8.3 minutes, over the requests whose answers were kept
Latency: the time from sending one request to receiving its answer, measured from our computer, network and any retries within the request included. Request time: the time at least one request was in flight, over the requests whose answers were kept. It is not the run's elapsed time: it leaves out requests that failed and any pause before a run was resumed.
Cost
- $ per 1,000 questions
- $0.029
- Whole run
- $0.063
- Basis
- Estimated from published prices and recorded token usage
- Price
- $0.042 input, output free per 1M tokens (source, 2026-10-16)
- Input tokens per question
- 687
- Output tokens per question
- 52
Going first in front of an LLM
Replay of recorded answers, cutoff chosen on half the questions, averaged over the splits. Savings are estimates from API prices; the Method page explains how.
| LLM behind it | Share taken | Accuracy change | Saving |
|---|---|---|---|
| Gemini 3.8 Flashhigh reasoning · 16 questions per request | 11.8% | −0.09 points | 11% |
| GPT-6 Astrahigh reasoning · 16 questions per request | 11.8% | −0.09 points | 11% |
| GPT-6.1 Solhigh reasoning · 16 questions per request | 11.8% | −0.09 points | 10% |
| GPT-6 Astralow reasoning · 16 questions per request | 11.8% | −0.09 points | 11% |
| GPT-6 Astralow reasoning · 1 question per request | 11.8% | −0.09 points | 12% |
| Gemini 3.8 Flashlow reasoning · 1 question per request | 13.6% | −0.13 points | 11% |
| GPT-6.1 Sollow reasoning · 1 question per request | 13.7% | −0.13 points | 13% |
| GPT-6.1 Sollow reasoning · 16 questions per request | 11.8% | −0.09 points | 9% |
| Claude Opus 5.5high reasoning · 16 questions per request | 14.5% | −0.14 points | 14% |
| GPT-6.1 Terrahigh reasoning · 16 questions per request | 12.0% | −0.10 points | 6% |
| Claude Opus 5.5low reasoning · 1 question per request | 13.4% | −0.11 points | 13% |
| Gemini 3.8 Flashlow reasoning · 16 questions per request | 15.5% | −0.14 points | 5% |
| GPT-6.1 Terralow reasoning · 16 questions per request | 14.3% | −0.17 points | 7% |
| Claude Opus 5.5low reasoning · 16 questions per request | 15.9% | −0.15 points | 15% |
| Claude Sonnet 5.5high reasoning · 1 question per request | 12.4% | −0.12 points | 12% |
| Claude Sonnet 5.5high reasoning · 16 questions per request | 29.2% | −0.12 points | 28% |