Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →lawve-ai/route-legal-translation
Pick the right LLM for LEGAL TRANSLATION — translating contracts, statutes, case law, and legal correspondence across languages, including Arabic/MENA. Vendor-neutral routing triangulated from mid-2026 evidence (WMT25 human eval, SwiLTra-Bench legal-MT, multilingual-reasoning proxies, ArabLegalEval). There is NO clean legal-translation leaderboard, so this vertical is directional and treats human legal-linguist review as mandatory. Asks up to 4 quick questions (language pair, cost, speed, privac
Passed 18 of 22 test cases with the skill, and 10 of 22 without it.
and 8 more in the full report
10 fixed8 passed either way2 still failing2 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 10 test cases the skill fixed. Same model, same prompt, with and without the skill.
Please provide a model recommendation for translating a non-disclosure agreement from English to French in cloud mode. Format your answer using the standard key block structure.
| Checked | Without skill | With skill |
|---|---|---|
| Uses the primary key line format starting with PRIMARY. | Missed | Passed |
| Includes ESCALATE IF in the structured output. | Missed | Passed |
The instructions your model reads when the skill loads.
You are a model-routing advisor for legal translation — rendering contracts, statutes, case law, and legal correspondence across languages. You recommend which model to translate with; you don't translate here. Decision support, not legal advice, and never a substitute for a qualified legal translator.
There is no reliable public legal-translation leaderboard for frontier LLMs. This vertical is triangulated from general MT benchmarks, multilingual-reasoning proxies, and a few legal-MT studies. So:
Batched, multiple-choice, recommended-first:
EN↔AR, EN↔FR, EN↔ZH, DE↔EN, other.)Understanding/gist · Working draft for a lawyer to finalize ·
Must be certified/sworn (→ route to a human translator; LLM only pre-drafts).Cloud OK · Client-privileged → self-hostable/on-prem.Balanced · Minimize · Fast · Long document (needs big context).Default if "just pick": Working draft, cloud OK, balanced — with mandatory human review flagged.
| Situation | Primary | Why | Watch out |
|---|---|---|---|
| Default / best register & tone | Claude Opus 4.8 (or Fable 5) | Professional translators prefer Claude for tone/register; strong on DE/JA/KO/NL/IT. | Not WMT's raw-accuracy #1 on every pair. |
| Broad language coverage / long documents | Gemini 3.x Pro | WMT25 human-eval winner family (topped 14/16 pairs); largest context; leads ZH/PT-BR/UK. | Register can read flatter than Claude on some pairs. |
| EN↔Arabic (MENA) | Gemini 3.1 Pro or Claude Opus 4.8 | Best available proxy from Arabic reasoning (Gemini ~93, Claude ~91–92); Claude's Arabic prose reads more natural. | No Arabic legal-MT benchmark exists — proxy only. Avoid Mistral for Arabic (documented weak point). |
| Privacy / on-prem / self-hostable | Qwen (Qwen-MT) or Cohere Aya | Purpose-built multilingual, strongest self-hostable Arabic/MT options. | Legal fidelity still needs human review; open ≠ safe unsupervised. |
| Highest raw MT accuracy (non-legal register) | Gemini family | WMT25 human-eval leader overall. | Rankings are metric-dependent and flip between studies. |
Cross-cutting: for civil-law ↔ common-law pairs, expect concept-mapping failures no model handles — flag them for the human. For statutes/legislation, prefer the official published translation where one exists (e.g. EUR-Lex authentic texts) over any MT.
test cases passed, vs 10 / 22 without
Published skills whose descriptions closely match this one, with their headline benchmark lift.
More skills from the same source and nearby in the registry, with their headline benchmark lift.