GDPR DPA Article 28 Validation
Given a data-processing addendum and an Article 28 requirements checklist, classify each requirement as present, missing, or ambiguous with supporting text spans.
View token rates
Compare 2 router rates · observed 9/11/2026
Compare 2 router rates · observed 9/11/2026
All 12 model results and methodology
Generated 100/100 examples
| Model | Tier | Quality | Judged | Scenario cost |
|---|---|---|---|---|
| Nemotron Nano 9B v2 | small | 24% | 100/100 | $28.07 |
| Qwen3 235B A22B | mid | 26% | 100/100 | $97.46 |
| DeepSeek V3 | mid | 19% | 100/100 | $53.30 |
| Mistral Large 2407 | mid | 19% | 100/100 | $125 |
| Arcee Trinity Large Thinking | mid | 24% | 100/100 | $138 |
| GPT-5.4 | frontier | 70% | 100/100 | $1112 |
| Claude Opus 4.7 | frontier | 84% | 100/100 | $1904 |
| Gemini 3.1 Pro Preview | frontier | 43% | 100/100 | $2167 |
| Gemma 4 E4B IT | small | 14% | 100/100 | $3.66 |
| Granite 4.1 8B | small | 19% | 100/100 | $7.41 |
| Ministral 8B Instruct 2410 | small | 12% | 100/100 | $18.69 |
| Qwen3 4B Instruct 2507 | small | 20% | 100/100 | $53.55 |
LLM-judge pass rate on 100 synthetic examples. Generator: gpt-5.2. Judge: gpt-5.2. Evaluated 2026-09-10T09:02:02.487Z.
Directional: measured on a synthetic eval set generated by drydock. Cost/latency are not yet captured for taskrouter-run benchmarks.
drydock