No direct translation benchmark is included in the current Artificial Analysis LLM endpoint.
No trustworthy winner is published until comparable benchmark data can be matched to these models.
See the benchmark research below.
Trusted translation benchmarks exist, but their published systems, language pairs, and evaluation setups do not yet map cleanly to the commercial LLM rows in this database. Naming a winner now would mix incompatible results.
Easy Benchmarks will publish a translation winner only when one source provides comparable scores for identifiable models across the same language pairs, prompts, and evaluation method.