If you want one API that can handle all three well, my practical recommendation is:

Best overall: DeepL API

• Short UI strings: Very strong quality and natural phrasing.
• Long documents: Solid, especially for high-quality translation workflows.
• Long-tail languages: Good coverage, though not the broadest.
• Strengths: Often best-in-class translation quality for major European languages.
• Tradeoff: Language coverage is narrower than Google/Microsoft, and it can be pricier depending on usage.

Best for widest language coverage: Google Cloud Translation

• Short UI strings: Good, especially when you need scale and many locales.
• Long documents: Handles large volumes well.
• Long-tail languages: Usually the strongest option if you need many rare or less common languages.
• Strengths: Broadest language support and robust infrastructure.
• Tradeoff: Quality can be less polished than DeepL for some language pairs.

Good enterprise alternative: Microsoft Translator

• Short UI strings: Reliable.
• Long documents: Fine for batch/document workflows.
• Long-tail languages: Decent coverage, sometimes better than expected.
• Strengths: Often cost-effective and easy to integrate in Microsoft-heavy environments.
• Tradeoff: Quality is generally competitive, but not usually the top choice for premium wording.

My recommendation by use case

• Mostly short UI strings, quality matters most: DeepL
• Need broadest language coverage, including long-tail languages: Google Cloud Translation
• Need a balanced enterprise option or already use Azure: Microsoft Translator

Best practical strategy

If your product needs short UI strings + long documents + long-tail languages, I’d usually suggest:

1. Primary: Google Cloud Translation for coverage
2. Secondary/premium path: DeepL for high-value UI text or premium locales
3. Fallback: Microsoft Translator if you want another broad-coverage backup

If you want, I can also give you a decision matrix comparing them on quality, cost, latency, glossary support, and language coverage.