Define your own test tasks
Create tasks with instructions, documents and pass/fail criteria. Import existing task collections when you want a starting point.
LegalBenchmarks by Eigenwelt Labs
Your tasks. Your standards.
Your model evaluation.
Evaluate connected models with your own documents, tasks and criteria. Compare runs, inspect individual results and test what changes when you adjust tools, Skills or workflows.
Part of LegalWork. Available in the same app.
LegalBenchmarks
Evaluate the work that matters to you.
Compare models on the same work
Choose the models to test and a judge model to grade the outputs. Compare results across the same tasks and criteria.
Inspect individual results
Open a result to review criterion scores, judge feedback, produced documents and the underlying conversation.
Test what makes a difference
Compare runs with selected tools, Skills or workflows disabled. See how each change affects performance on your tasks.
Pricing
LegalWork app
Free
Including commercial use.
Local models or your own API keys.
Download LegalWork↓- Complete MIT-licensed desktop app
- Your own providers or local models
- Document editors and Office add-ins
- Local Tasks, Workflows, Recorder and Evaluations
- Inspect, adapt and run the code yourself
Eigenwelt Sync
€15 / month
Per seat · €180 billed yearly
All collaboration features, with your own AI.
Start your 7-day trial↗- No AI usage included. Add credits whenever you need them.
- Bring your own AI subscription or API key. Sign in with your OpenAI account and use your ChatGPT plan.
- Team sync for Tasks, Workflows, Skills, integrations and settings
- Admin dashboard: connect storage for everyone, assign usage
- Priority support
Eigenwelt Plus
€29 / month
Per seat · €348 billed yearly
Best for firms starting with AI.
Start your 7-day trial↗- Everything included in Sync
- €30 of AI usage included per seat each month
- Premium models: GPT-6.1 Sol, Claude Opus 5.5 and Gemini 3.8 Flash
- Chat, drafting, review and redlining
- Intake for emails and API requests
- As admin, decide which models your team can use
- Hosting in the EU or the US, zero data retention, never used for training
Prices exclude VAT. Trial for your first eligible subscription; payment method required. Service terms ↗