Language Support Update: 12 New Languages Now Available
When we launched AI humanization, we supported English. That made sense for our initial user base. But as more international customers started using the platform, we kept getting the same request: “Can you humanize content in our language too?”
We’ve been building support for additional languages in parallel with our core features. We’re excited to announce that we now support 12 new languages in addition to English.
What’s New
Humanization now supports: Spanish, French, German, Italian, Portuguese, Dutch, Swedish, Norwegian, Danish, Polish, Czech, and Japanese.
Each language went through the same rigorous training process as English. We worked with native speakers to understand cultural communication patterns, brand voice considerations, and the specific challenges of humanizing content in that language. The result is that humanization works as well in Spanish as it does in English, which is to say, really well.
How We Built This
Language expansion requires more than just translation. We needed to understand how each language naturally communicates. What makes content sound human in German is different from what makes it sound human in Spanish. Sentence structure, formality levels, humor, emotional expression, they all vary.
We trained our models on large datasets of human-written content in each language, focusing on the kinds of content our customers create: marketing copy, product descriptions, emails, blog posts. That context-specific training means the humanization understands what “natural” looks like in your industry and your language.
We also worked with fluent speakers who gave us feedback on draft outputs. That human feedback was crucial. Machine learning gets you 80% of the way there; human expertise gets you the final 20%.
Feature Parity Across Languages
You get all the same features in every supported language. Tone Matching works. Batch processing works. Custom tone profiles work. The Batch Endpoint processes multiple languages in a single request if you want.
That consistency matters. If you manage content in multiple languages, you have one platform that works the same way across all of them.
API Implementation
Using the new languages is simple. Include a language parameter with your API request. Options are: en, es, fr, de, it, pt, nl, sv, no, da, pl, cs, ja. If you don’t specify a language, we default to English.
For batch processing, you can mix languages in a single batch. You might humanize some English product descriptions and some Spanish product descriptions in one request. That’s useful for international e-commerce teams managing multilingual catalogs.
Check out our API documentation for implementation examples in every supported language.
Pricing
Language support doesn’t change your pricing. You pay the same per-token rate regardless of which language you’re using. That keeps costs predictable for international teams.
What This Means for International Teams
If you manage content in multiple languages, you can now humanize all of it on a single platform. No more switching between different tools or outsourcing humanization of non-English content. One account, consistent results across languages.
We’re seeing early use of this from multinational companies managing brand voice across language boundaries, international content agencies serving multiple markets, and global e-commerce teams.
More Languages Coming
We’re planning to add more languages in the coming months. Chinese, Korean, Russian, and Arabic are on our roadmap. If you have a specific language request, reach out, it helps us prioritize development.
The same principle applies to all new languages: native speaker expertise, context-specific training, feature parity with existing languages.
Get Started With New Languages
If you’ve been waiting for humanization support in a language beyond English, you can start using it today. Visit our pricing page to upgrade your plan if needed, or check out our documentation to get implementation details.
Want to explore the new languages before committing? Try them in our interface with sample content and see how well humanization works for your use case.
One platform, multiple languages, consistent results. That’s what we built.
The full list of supported languages
The AI Humanizer API now supports 50+ languages across major writing systems. Here’s how the language coverage breaks down by typical use case.
Highest-quality output (5+ years of training data)
- English – US, UK, AU, IN dialect handling. The reference language for most of our humanization research.
- Spanish – Iberian and Latin American variants supported with appropriate idiom selection.
- French – Standard, Canadian, and African dialect awareness.
- German – Du/Sie register handling, formal and informal output.
- Mandarin Chinese – Simplified and Traditional, with mainland vs. Taiwanese phrasing.
- Portuguese – Brazilian and European variants, with appropriate vocabulary.
- Italian, Dutch, Russian, Japanese, Korean, Arabic – full feature parity with English.
Strong support (continuously improving)
Hindi, Bengali, Turkish, Polish, Vietnamese, Thai, Indonesian, Malay, Czech, Greek, Hebrew, Swedish, Norwegian, Danish, Finnish, Romanian, Hungarian, and Ukrainian – these languages cover the bulk of global content needs and produce reliable humanization with the same API surface.
Good support (works for most content types)
Persian, Urdu, Tamil, Telugu, Marathi, Gujarati, Punjabi, Bulgarian, Croatian, Serbian, Slovak, Slovenian, Lithuanian, Latvian, Estonian, Tagalog, Swahili, Afrikaans. We continuously add training signal for these languages and recommend running a small batch through your detector of choice before scaling.
How to specify language in your API request
Pass the ISO 639-1 code in the language parameter. The API auto-detects when omitted, but explicit language selection is faster and more accurate, especially for shorter inputs:
POST /v1/humanize
{
"text": "El uso de la inteligencia artificial...",
"tone": "professional",
"language": "es"
}For mixed-language documents (e.g. English with technical terms in another language), our engine handles code-switching automatically – no special parameter required. See the full API documentation for the complete parameter reference.
Tone selection across languages
The four supported tones – professional, conversational, casual, and academic – translate differently across languages. In Japanese, for example, “casual” output uses plain forms (です/ます-less), while “professional” output applies appropriate keigo. In German, the engine adjusts du/Sie automatically based on tone choice.
If you’re generating a multi-language campaign, send the same source text in each language with the same tone parameter – the engine will produce equivalent register across all variants. This is especially useful for global content marketing teams running coordinated rollouts.
Performance benchmarks per language
Latency is consistent across languages – the average response stays under 2 seconds for inputs up to 1,000 words regardless of language. Output quality (measured by AI-detection bypass rate against major detectors) is highest for the top tier, with the 8 tier-1 languages achieving consistent results across our internal test suite.
For high-volume multi-language pipelines, the batch endpoint accepts up to 100 items per request and processes them in parallel. You can mix languages in a single batch – the API detects each one independently.
Common multi-language workflows
Localizing AI-drafted content
Many teams generate content in English first (where their AI tools are strongest), then translate to target markets. The translation often produces stiff, mechanical-sounding output. Run the translated text through the AI Humanizer API in the target language to restore natural flow.
SEO content for international markets
If you’re publishing keyword-targeted content in multiple languages, run each language version through the API individually with the same tone preset. Search engines reward natural-sounding regional content, and humanization improves dwell time and engagement metrics – which are increasingly important ranking factors.
Customer support and documentation
Help articles, knowledge bases, and chatbot responses translated from a primary language often read robotic to native speakers. Plug the AI Humanizer API into your translation pipeline to soften the output. See our chatbot use case for a worked example.
What’s next for language support
We’re actively expanding to additional languages this quarter, with current focus on smaller European languages (Catalan, Basque, Galician) and several African languages (Hausa, Igbo, Yoruba, Amharic). If your team needs a specific language, let us know – language requests from paying customers are prioritized in the roadmap.
Frequently asked questions
Does humanization quality vary by language?
Yes. The 12 tier-1 languages produce the most consistent results because they have the largest training corpora. Tier-2 and tier-3 languages still produce strong output but may show slightly more variance run-to-run. We recommend testing your specific use case before scaling.
Can I use the API for languages not on the supported list?
Yes – the engine will attempt humanization on any input. Output quality for unsupported languages is best-effort. If you have a specific need, contact us and we’ll evaluate adding it to the roadmap.
How do I detect the language automatically?
Omit the language parameter. The API will detect language from the input and humanize accordingly. For very short inputs (<50 characters), explicit language specification is more reliable.
Is pricing the same across all languages?
Yes. Word count is calculated consistently across languages, and our pricing applies uniformly regardless of which language you’re processing.
Get started with multi-language humanization
Sign up for a free API key (10,000 words per month, no credit card) and try humanization in your target languages. The same endpoint and the same parameter shape work for all 50+ languages – pick a tone, optionally specify the language, and the engine handles the rest.
