The State of Translation Automation 2026 · Languages
What it takes to automate translation from English into Haitian Creole against 10 requirements in three levels: the solutions that meet them, what they cost and how fast they are, from the 2026 benchmark of 57 AI models and MT engines.
A solution is a model plus its harness: prompts, glossaries and other context, pre- and post-processing. It meets a quality level when it meets every requirement of the level.
| Requirement | Description | Met when |
|---|---|---|
| Basic requirements for any company | ||
| General translation quality | Translation errors of every type, such as mistranslations, omissions and grammar errors, classified by type and severity following MQM, the industry standard. | No critical error, at most 3 major errors in the four test texts (3,034 words) and at most 20 errors per 1,000 words, weighted by severity (minor 1, major 5, critical 10). |
| Standard requirements adds the rules companies set for their content | ||
| Terminology | The approved translation of every glossary term, in any of its correct forms. | 100% of the glossary terms. We test it in the 16 languages with a glossary, so it does not apply in Haitian Creole. |
| Tone of voice | The formal register in every sentence in the second person or the imperative. | 100% of those sentences. |
| Tags | Every HTML tag of the source, with its attributes, in the translation. | 100% of the tags. |
| Unicode character integrity | Every character intact, with no corrupted characters and no letter from another script that only looks like the right one. | 100% of the characters. |
| Locale conventions | The punctuation marks of the target language in place of the English ones. | No pass-or-fail bar. We report it as a share, because whether to localize punctuation is a matter of house style. |
| HTML entities | No HTML character code, such as the code for a quotation mark, that the source does not carry. | No added HTML entity. |
| Valid JSON | The translated JSON document still parses. | The document parses. |
| Spacing at tag boundaries | The spaces next to HTML tags stay as in the source. | 100% of the tag boundaries. |
| Advanced requirements adds a requirement for some companies and use cases | ||
| Gender neutrality | A gender-neutral translation wherever the English source leaves the gender of a person open, as in “your manager” or “the expert”. | 100% of the sentences that refer to a person. |
| Metric pair | Metrics | Agreement in Haitian Creole | 90% band | Minimum | Result |
|---|---|---|---|---|---|
| Similarity | LaBSE, Cohere embed v3 | 0.415 | 0.399 to 0.432 | 0.461 | Under the minimum |
| Quality estimation | COMET-Kiwi, MetricX | — | — | 0.534 | Cannot be applied |
MetricX, the other metric of the quality-estimation pair, covers Haitian Creole. The page also reports which models support Haitian Creole, the tokens per character and the length of a translation, and the tone of voice and gender neutrality requirements. None of these are results on a requirement.
Pick a quality level, or your own requirements, bars and constraints, to see the shortlisted solutions that meet them in Haitian Creole, with a column for each requirement you select. Under Optimize for, choose which of them comes first: the one with the lowest cost or the shortest time. Click a column heading to sort by it.
We did not shortlist solutions in Haitian Creole, so the explorer has none: why.
A time limit of 15 seconds per 1,000 words leaves 9 models: the 7 real-time ones, Yandex Cloud Translate v2 at 3.0 seconds per 1,000 words, and OpenAI GPT-5.6 Luna. A limit of 35 seconds per 1,000 words also leaves xAI Grok 4.20 (non-reasoning), OpenAI GPT-4.1 and OpenAI GPT-5.6 Terra. Translations from these three LLMs and OpenAI GPT-5.6 Luna have 0.31 to 0.343 tokens per character, below the median of 0.36. With the next LLM, Anthropic Claude Haiku 4.5, a translation takes 41 seconds per 1,000 words.
Translations from Google Gemma 4 cost $0.13 per million characters but take 196 seconds per 1,000 words, and those from Yandex Cloud Translate v2 take 3.0 seconds per 1,000 words at $3.94 per million characters.
We cannot assess general translation quality in Haitian Creole: how confident we are.
We cannot assess general translation quality in Haitian Creole: how confident we are.
We cannot assess general translation quality in Haitian Creole: how confident we are.
| Model | US dollars per million characters | Class |
|---|---|---|
| Google Gemma 4 | $0.13 | Open-weight LLMs |
| OpenAI GPT-5.6 Luna | $0.41 | LLMs |
| MiniMax M2.5 | $0.59 | LLMs |
| Mistral Large 3 | $0.72 | LLMs |
| DeepSeek V3.2 | $0.89 | LLMs |
| Mistral Medium 3.1 | $0.89 | LLMs |
| xAI Grok 4.20 (non-reasoning) | $1.1 | LLMs |
| xAI Grok 4.20 (reasoning) | $1.1 | LLMs |
| Alibaba Qwen3-235B-A22B | $1.14 | Open-weight LLMs |
| MoonshotAI Kimi K2.5 | $1.17 | LLMs |
| z.ai GLM-5 | $1.84 | LLMs |
| Anthropic Claude Haiku 4.5 | $2.13 | LLMs |
| OpenAI GPT-4.1 | $2.83 | LLMs |
| Google Gemini 3.5 Flash | $3.38 | LLMs |
| AI21 Jamba Large 1.7 | $3.43 | LLMs |
| Yandex Cloud Translate v2 | $3.94 | MT engines |
| OpenAI GPT-5.6 Terra | $4.11 | LLMs |
| Google Gemini 3.1 Pro Preview | $4.54 | LLMs |
| Anthropic Claude Sonnet 5 | $5.76 | LLMs |
| Anthropic Claude Sonnet 4.6 | $6.53 | LLMs |
| Youdao Translation Service API | $6.71 | MT engines |
| OpenAI GPT-5.6 Sol | $7.07 | LLMs |
| Microsoft Translator | $10 | MT engines |
| OpenAI GPT-5.5 | $10.26 | LLMs |
| Anthropic Claude Opus 4.8 | $14.35 | LLMs |
| Amazon Translate | $15 | MT engines |
| Google Cloud Advanced Translation API | $20 | MT engines |
| Translated Lara 2 | $24.99 | LLM-based MT engines |
| DeepL Next Gen | $27.5 | LLM-based MT engines |
| Anthropic Claude Fable 5 | $28.94 | LLMs |
| Model | Tokens per character | Class |
|---|---|---|
| Anthropic Claude Opus 4.8 | 0.526 | LLMs |
| Anthropic Claude Sonnet 5 | 0.525 | LLMs |
| Anthropic Claude Fable 5 | 0.524 | LLMs |
| Mistral Medium 3.1 | 0.42 | LLMs |
| Mistral Large 3 | 0.415 | LLMs |
| Anthropic Claude Sonnet 4.6 | 0.406 | LLMs |
| Anthropic Claude Haiku 4.5 | 0.404 | LLMs |
| AI21 Jamba Large 1.7 | 0.398 | LLMs |
| Alibaba Qwen3-235B-A22B | 0.378 | Open-weight LLMs |
| DeepSeek V3.2 | 0.368 | LLMs |
| z.ai GLM-5 | 0.361 | LLMs |
| MoonshotAI Kimi K2.5 | 0.36 | LLMs |
| Cohere Command A Translate | 0.359 | LLMs |
| xAI Grok 4.20 (non-reasoning) | 0.343 | LLMs |
| xAI Grok 4.20 (reasoning) | 0.342 | LLMs |
| Google Gemini 3.1 Pro Preview | 0.341 | LLMs |
| Google Gemini 3.5 Flash | 0.34 | LLMs |
| Google Gemma 4 | 0.339 | Open-weight LLMs |
| OpenAI GPT-4.1 | 0.313 | LLMs |
| OpenAI GPT-5.5 | 0.311 | LLMs |
| OpenAI GPT-5.6 Luna | 0.311 | LLMs |
| OpenAI GPT-5.6 Sol | 0.311 | LLMs |
| OpenAI GPT-5.6 Terra | 0.31 | LLMs |
| MiniMax M2.5 | 0.304 | LLMs |
Over the 24 models we counted tokens for, a translation has a median 0.36 tokens per character, 1.5 times as many as English text. Tokens per character do not set the cost alone. Translations from Anthropic Claude Haiku 4.5 and Anthropic Claude Sonnet 4.6 have 0.404 and 0.406 tokens per character and cost $2.13 and $6.53 per million characters. Translations from Mistral Large 3 have more tokens per character, 0.415, and cost $0.72 per million characters.
| Model | Seconds per 1,000 words | Class |
|---|---|---|
| Google Cloud Advanced Translation API | 0.4 | MT engines |
| ModernMT Enterprise Edition | 0.4 | MT engines |
| Microsoft Translator | 0.5 | MT engines |
| DeepL Next Gen | 0.9 | LLM-based MT engines |
| Translated Lara 2 | 1.3 | LLM-based MT engines |
| Amazon Translate | 1.4 | MT engines |
| OpenAI gpt-oss-120b | 2 | Open-weight LLMs |
| Yandex Cloud Translate v2 | 3 | MT engines |
| OpenAI GPT-5.6 Luna | 14 | LLMs |
| xAI Grok 4.20 (non-reasoning) | 25 | LLMs |
| OpenAI GPT-4.1 | 33 | LLMs |
| OpenAI GPT-5.6 Terra | 33 | LLMs |
| Anthropic Claude Haiku 4.5 | 41 | LLMs |
| Swiss AI Apertus-8B-2509 | 42 | Open-weight LLMs |
| Mistral Medium 3.1 | 42 | LLMs |
| AI21 Jamba Large 1.7 | 46 | LLMs |
| MiniMax M2.5 | 50 | LLMs |
| Cohere Command A Translate | 54 | LLMs |
| Youdao Translation Service API | 55 | MT engines |
| Google Gemini 3.5 Flash | 57 | LLMs |
| xAI Grok 4.20 (reasoning) | 58 | LLMs |
| OpenAI GPT-5.6 Sol | 59 | LLMs |
| OpenAI GPT-5.5 | 65 | LLMs |
| DeepSeek V3.2 | 67 | LLMs |
| Anthropic Claude Sonnet 5 | 68 | LLMs |
| Mistral Large 3 | 70 | LLMs |
| Anthropic Claude Fable 5 | 76 | LLMs |
| Google Gemini 3.1 Pro Preview | 76 | LLMs |
| Anthropic Claude Sonnet 4.6 | 77 | LLMs |
| Anthropic Claude Opus 4.8 | 87 | LLMs |
| Alibaba Qwen3-235B-A22B | 135 | Open-weight LLMs |
| z.ai GLM-5 | 169 | LLMs |
| Google Gemma 4 | 196 | Open-weight LLMs |
| MoonshotAI Kimi K2.5 | 272 | LLMs |
Real-time models are those the shortlist counts as real-time, under 2 seconds per text across the languages we timed. The chart shows the time of each model into Haitian Creole.
Of the MT engines, only Yandex Cloud Translate v2 and Youdao Translation Service API are not real-time, and translations from Youdao Translation Service API take 55 seconds per 1,000 words. Among the LLMs, translations from xAI Grok 4.20 (non-reasoning) take 25 seconds per 1,000 words, and those from xAI Grok 4.20 (reasoning) 58. Translation times on the four open-weight LLMs range from 2.0 seconds per 1,000 words on OpenAI gpt-oss-120b to 196 on Google Gemma 4.
This page has no tone of voice or gender neutrality result for Haitian Creole.
Leave feedback