SaaS MSA Liability Redlining
Given a software master-services agreement and an approved clause policy, identify indemnity and limitation-of-liability deviations and return clause-level suggested edits.
View token rates · observed 9/11/2026
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 | 5% | 100/100 | $41.07 |
| Qwen3 235B A22B | mid | 37% | 100/100 | $258 |
| DeepSeek V3 | mid | 55% | 100/100 | $101 |
| Mistral Large 2407 | mid | 43% | 100/100 | $164 |
| Arcee Trinity Large Thinking | mid | 29% | 100/100 | $121 |
| GPT-5.4 | frontier | 73% | 100/100 | $2181 |
| Claude Opus 4.7 | frontier | 60% | 100/100 | $5082 |
| Gemini 3.1 Pro Preview | frontier | 23% | 100/100 | $2450 |
| Gemma 4 E4B IT | small | 28% | 100/100 | $7.41 |
| Granite 4.1 8B | small | 6% | 100/100 | $13.53 |
| Ministral 8B Instruct 2410 | small | 2% | 100/100 | $21.20 |
| Qwen3 4B Instruct 2507 | small | 4% | 100/100 | $76.82 |
LLM-judge pass rate on 100 synthetic examples. Generator: gpt-5.2. Judge: gpt-5.2. Evaluated 2026-09-10T09:02:02.524Z.
Directional: measured on a synthetic eval set generated by drydock. Cost/latency are not yet captured for taskrouter-run benchmarks.
drydock