Short answer: start with a fast, cost-efficient model. For most everyday website translation, you do not need the most expensive flagship or “frontier” model. Compare a small sample of your real content for meaning, tone, terminology, speed and cost. Choose through a governed Umbraco.AI profile so editors use a known model and prompt.
Is there one best AI model for translation?
No single model is best for every language pair, content type and organisation. A model that handles English-to-German product copy well may be less convincing for Japanese interface labels, Arabic right-to-left content or specialist legal wording. Models and prices also change, so a permanent recommendation based on one benchmark quickly becomes stale.
A useful selection process is repeatable: define what good translation means for the project, test a fixed content set, record the results and review the choice when models or requirements change.
What should you compare?
| Criterion | What to test | Why it matters |
|---|---|---|
| Meaning | Ambiguous phrases, references and calls to action | Fluent output can still change the intended claim |
| Language quality | Each real source and target pair | Capability varies by language and regional variant |
| Terminology | Brand names, protected terms and required translations | Consistency matters across pages and releases |
| Structure | Rich text, placeholders and short UI strings | Output must remain usable inside the CMS |
| Latency and reliability | Single pages and a representative batch | Bulk workflows amplify slow responses and rate limits |
| Cost | Input and output usage for a measured content sample | Model price alone does not reveal whole-site cost |
| Governance | Data region, retention, account controls and audit needs | Content may include confidential or personal information |
Build a realistic translation test set
Use a small, version-controlled sample containing the content editors actually publish:
- short headings and navigation labels;
- long-form editorial copy;
- rich text containing links and inline formatting;
- brand terminology and product names;
- numbers, dates and placeholders that must remain intact;
- ambiguous source phrases that require context;
- content from nested blocks; and
- every target language that materially affects the project.
Have a fluent reviewer score meaning, naturalness, terminology and the amount of editing required. The best model is usually the one that reduces review effort reliably—not necessarily the one that produces the most polished first sentence.
Do you need a frontier model for website translation?
Usually, no. Translating service pages, product descriptions, articles, FAQs and dictionary items is a focused language task. Modern efficient models can handle this content well without the cost of a provider's most powerful model. Start with a fast, lower-cost option and move up only when a comparison on your own content shows a worthwhile improvement.
Lower-cost models can perform very well with minimal reasoning. Where the model and installed provider expose effort controls, start low and increase effort only when your own comparison shows an improvement. Setting names and supported levels vary. The examples below are candidates to evaluate through their supported Umbraco.AI providers, rather than a claim that we have tested every model in Omni.
Fast, cost-efficient models to consider
| Provider and model | Why consider it? | Umbraco.AI connection |
|---|---|---|
| OpenAI GPT-6 Luna | OpenAI's efficient tier for focused, high-volume tasks; a sensible starting candidate for routine website translation. | OpenAI; select gpt-6-luna when available in your Chat profile. |
| Anthropic Claude Sonnet | A balance of speed and capability, at a lower token cost than Opus. A useful candidate for general site copy and marketing tone. | Anthropic; choose an available Sonnet model in your Chat profile. |
| Anthropic Claude Haiku | Anthropic's faster, lower-cost tier; worth comparing for dictionary items, short labels and straightforward page content. | Anthropic; choose an available Haiku model in your Chat profile. |
| Google Gemini Flash / Flash-Lite | Google's efficient model families, with Flash-Lite focused on low-cost throughput. Useful candidates for larger volumes of routine content. | Google; choose a Flash or Flash-Lite model supported by your installed provider version. |
Sonnet is a balanced tier rather than the cheapest option; Luna, Haiku and Flash-Lite are particularly worth considering when cost and throughput matter. Model availability depends on your provider account and installed Umbraco.AI integration. Configure the model as a Chat profile and translate one representative page before running a section. See provider setup and profile configuration.
Check current OpenAI, Anthropic and Google pricing when choosing a model. Compare total job cost and completion time, including any retries, rather than assuming a bigger model buys a better translation.
A larger model can still be useful for unusually ambiguous source material or nuanced campaign copy. Keep it as an optional profile for pages that benefit from it. Specialised legal, medical and regulated content deserves qualified human review, whichever model you choose.
How should language pairs affect your choice?
Start with an efficient model for each language pair and check the details that matter to your audience. A different script or regional variant does not automatically mean you need a premium model. Use the following checks to decide whether your default profile is already doing the job well.
| Language pair | What to watch for |
|---|---|
| English ↔ French, German, Spanish, Italian, Dutch, Portuguese | Formal/informal address (tu/vous, du/Sie) and compound nouns |
| English ↔ Nordic languages, Polish, Czech | Case endings and inflection after glossary substitution |
| English ↔ Japanese, Korean | Politeness level consistency across a page; word order in short headings |
| English ↔ Chinese (Simplified or Traditional) | Correct variant, punctuation width and spacing around inline markup |
| English ↔ Arabic, Hebrew | Right-to-left rendering in templates and mixed LTR fragments such as URLs |
| Any pair not involving English | Meaning, idioms and natural phrasing in both languages |
Right-to-left languages deserve particular care: the translation may be correct while the page still renders badly because the template lacks adir attribute. Verify the rendered page, not only the value in the backoffice. Costs by tier are set out in estimating AI translation costs.
Use profiles to govern the choice
Umbraco.AI chat profiles combine a provider, model and instruction set. A site can maintain a dependable default profile and, when useful, permit alternatives for faster drafts, higher-quality review work or a specific language pair.
How should the system prompt be written?
Keep the system prompt clear and narrow. State that the task is translation, require the original meaning and protected structure to remain intact, identify the intended audience or tone, and tell the model not to add commentary. Put stable terminology in glossary rules rather than growing one unmanageable prompt.
Common questions
Can editors choose different models?
Yes, when administrators intentionally allow multiple profiles. A smaller approved list is easier to support and produces more predictable results than unrestricted provider access.
Should the cheapest model translate the whole site?
Often, yes: a fast, lower-cost model can be a practical default for most of the site. Check a representative sample first, then use a different profile only for pages or language pairs where it makes a meaningful difference.
Will a frontier model always give a better translation?
No. A higher price buys broader capability, but routine translation may not need it. Clear instructions, useful context and consistent glossary rules can matter more than moving to the most expensive tier.
Should sensitive content be sent to an AI provider?
Not by default. Review the provider agreement, retention controls, data location and organisational policy, then exclude content that should not leave the application boundary.
Configure and test a provider
Follow the detailed Umbraco.AI configuration guide, review AI profiles and prompts, and use the safe AI translation workflow for the first production trial.