Extract term, renewal, notice, liability cap and governing law from service contracts with amendments
Exact prompt
# Extract key terms from a stack of service contracts Input: `contracts/` holds 5-8 plain-text service agreements (synthetic). Some end with one or more amendments. For each contract extract: - term_months: the initial term in months (a term stated in years counts 12 months per year). - auto_renew: true if the agreement renews automatically unless notice is given, false otherwise. - notice_days: days of notice needed to terminate or prevent renewal. - liability_cap_usd: the liability cap in US dollars. When the cap is defined by fees (for example "fees paid in the twelve (12) months preceding the claim"), compute it as monthly_fee_usd x that number of months. - governing_law: the state named in the governing-law clause, as written (e.g. "Delaware"). - Amendments at the end of a contract override the original clause they amend; apply them in order. - Numbers may be written in words, in digits, or both (e.g. "sixty (60) days"). Write `out/contracts.json`: `{"<contract file name without .txt>": {"term_months": <int>, "auto_renew": <bool>, "notice_days": <int>, "liability_cap_usd": <number>, "governing_law": "<state>"}, ...}`. ## How this run works - Work on your own: nobody will answer questions in this task, and writing to `ask/` counts against the run. - Write only under `out/`. Create `out/` if it is missing. Inputs are synthetic; don't fetch anything from the network. - Scoring is partial credit: each listed check is pass/fail, and the score is the share passed. Numbers are compared to within 0.005 unless the task says otherwise.
Read 5-8 synthetic service agreements and extract five commercial terms from each. Numbers appear as words, digits or both, some liability caps are defined as a number of months of fees, and signed amendments at the end override earlier clauses. Distractors include other day counts, percentages and a second state named in the notices clause. Original benchmark task written for usecaserank.com (not based on a specific public post). Headless, offline run on a reproducible fixture. Each run's notes say how its cost was measured (estimates are marked).
Success criteria: term_months auto_renew notice_days liability_cap governing_law amendments_applied every_contract_complete
Best model for this task so far: Claude Haiku 5.5 (effort default (no --effort flag) · thinking on) (success, $0.0048, 15s).
All runs (4)
75% pass rate across 4 eligible runs: 3 pass, 1 partial, 0 fail. Partial and failed runs are listed below with everything else.ⓘ
| Model | Cost | Time | Step-insⓘ | Autonomy | Outcome | Trustⓘ | Date | Compare |
|---|---|---|---|---|---|---|---|---|
| MMistral Small 3.2 24B Instruct (Venice)thinking on | $0.0012 | 16s | 0 | Fully autonomous | partial | Trust 80% | 2026-10-28 | |
| AClaude Opus 5.5default · thinking on | $0.13 | 18s | 0 | Fully autonomous | pass | Trust 80% | 2026-10-20 | |
| AClaude Haiku 5.5default · thinking on | $0.0048 | 15s | 0 | Fully autonomous | pass | Trust 80% | 2026-10-20 | |
| AClaude Sonnet 5.5default · thinking on | $0.05 | 9s | 0 | Fully autonomous | pass | Trust 80% | 2026-10-20 |
Embed this result
A live badge for this use case. It links back here and updates as new runs are published.
Get new runs and ranking changes weekly
Which models got better or cheaper on real tasks like this one.