# Jev 1.13 on bench-v4

Generated by `python3 -m decision_bench publish`. Do not edit by hand.

- Model id: `jev-1.13` (provider `typesafe`, model name sent: `jev-1.13.0`)
- Corpus: 1.0 (sha256 `3599baea0d9c7e4e...`)
- Harness commit: `4b919ae6f2c97e69dfe05f378c495e8604df994e`
- Run completed: 2026-09-23T13:50:29.093424+00:00

**Accuracy 92.4%** (990/1071; Wilson 95% 90.7%-93.9%). Rows with no valid answer: 0.

Cost $0.0463 (coverage 100%; basis price_table_estimate). Median latency 439 ms.

| Category | Correct | Accuracy | Wilson 95% |
| --- | --- | --- | --- |
| AI agents & evals | 135/154 | 87.7% | 81.5%-92.0% |
| Sales & commerce | 84/90 | 93.3% | 86.2%-96.9% |
| Data & analytics | 124/126 | 98.4% | 94.4%-99.6% |
| Design | 14/30 | 46.7% | 30.2%-63.9% |
| Documents & meetings | 76/87 | 87.4% | 78.8%-92.8% |
| Engineering | 146/150 | 97.3% | 93.3%-99.0% |
| Finance | 120/121 | 99.2% | 95.5%-99.9% |
| Legal | 121/126 | 96.0% | 91.0%-98.3% |
| Product management | 47/60 | 78.3% | 66.4%-86.9% |
| Trust & safety | 58/60 | 96.7% | 88.6%-99.1% |
| Customer support | 65/67 | 97.0% | 89.8%-99.2% |

| Task | Correct | Accuracy |
| --- | --- | --- |
| AGT-1 Which tool to call | 35/36 | 97.2% |
| AGT-2 Did the injection succeed? | 30/30 | 100.0% |
| AGT-3 Did the agent complete the task? | 23/30 | 76.7% |
| AGT-4 Is the answer backed by sources? | 21/30 | 70.0% |
| AGT-5 Which reply is better? | 26/28 | 92.9% |
| COM-1 Does this product match the query? | 24/30 | 80.0% |
| COM-2 What type of product is this? | 30/30 | 100.0% |
| COM-3 Which industry is this company in? | 30/30 | 100.0% |
| DAT-1 Which SQL answers the question? | 30/30 | 100.0% |
| DAT-2 Which cell answers the question? | 29/30 | 96.7% |
| DAT-3 Does the chart support the claim? | 32/32 | 100.0% |
| DAT-4 What does this column hold? | 33/34 | 97.1% |
| DOC-1 Where is this page from? | 30/30 | 100.0% |
| DOC-2 Decision, action item, or neither? | 21/30 | 70.0% |
| DOC-3 Which summary is this meeting's? | 25/27 | 92.6% |
| DSN-1 What does this icon mean? | 14/30 | 46.7% |
| ENG-1 Which file must change? | 30/30 | 100.0% |
| ENG-2 What kind of change is this? | 28/30 | 93.3% |
| ENG-3 Which weakness class is this? | 29/30 | 96.7% |
| ENG-4 Live secret or placeholder? | 29/30 | 96.7% |
| ENG-5 Which message describes this diff? | 30/30 | 100.0% |
| FIN-1 Which part of the 10-K? | 29/30 | 96.7% |
| FIN-2 What does this 8-K report? | 30/30 | 100.0% |
| FIN-3 Which figure is the total? | 30/30 | 100.0% |
| FIN-4 What does this number report? | 31/31 | 100.0% |
| LEG-1 Does the NDA say this? | 28/33 | 84.8% |
| LEG-2 What kind of clause is this? | 30/30 | 100.0% |
| LEG-3 Is this term unfair, and how? | 33/33 | 100.0% |
| LEG-4 Does the contract cover this? | 30/30 | 100.0% |
| PRD-1 Which headline won? | 17/30 | 56.7% |
| PRD-2 Major, minor or patch? | 30/30 | 100.0% |
| SAF-1 Attack on the assistant | 29/30 | 96.7% |
| SAF-2 Toxic comment or not | 29/30 | 96.7% |
| SUP-1 Which team owns this complaint? | 35/35 | 100.0% |
| SUP-2 What is the complaint about? | 30/32 | 93.8% |