The State of Translation Automation 2026 · Languages
What it takes to automate translation from English into Lao 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 Lao. |
| Tone of voice | The formal register (respectful vocabulary) 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. |
Pick a quality level, or your own requirements, bars and constraints, to see the shortlisted solutions that meet them in Lao, with a column for each requirement you select. Under Optimize for, choose which of them comes first: the one with the fewest total errors, the lowest cost or the shortest time. Click a column heading to sort by it.
The shortlisted solutions on 11 models, with their Intento LQA results. Total errors per 1,000 words add the weighted general translation quality errors (minor 1, major 5, critical 10) to the missed glossary terms, sentences in the wrong register, tag errors and sentences that fail gender neutrality, each counted once. Formatting counts as met or missed and does not add to the total errors. The models that did not make the shortlist have automatic scores only: the shortlist. Inconclusive: the three LQA judges agree on fewer than 90% of the sentences of the solution, so it has no general translation quality result (09-02). Gender neutrality: the Advanced requirements count the two best solutions of the language, with the gender-neutrality agent where they need it. The results of the other solutions are shown for information.
Neither of the two-step solutions (translation + post-edit) that meet every Standard requirement in Lao leads on errors, cost and time at once. The one on OpenAI GPT-5.6 Sol has 16.5 total errors per 1,000 words and costs $27.98 per million characters, against 18.1 and $28.72 for the one on Google Gemini 3.1 Pro Preview. It takes 220 seconds per 1,000 words, against 163. Optimizing for quality or cost selects OpenAI GPT-5.6 Sol, and optimizing for latency Google Gemini 3.1 Pro Preview.
The Advanced requirements add gender neutrality, and both two-step solutions (translation + post-edit) that meet every Standard requirement in Lao meet it without the gender-neutrality agent. So the Advanced requirements add nothing in cost or time here. The solution on OpenAI GPT-5.6 Sol still costs $27.98 per million characters, and the one on Google Gemini 3.1 Pro Preview still takes 163 seconds per 1,000 words.
Google Gemini 3.1 Pro Preview
one-step LLM adjusted for Standard requirements
Google Gemini 3.1 Pro Preview
one-step LLM adjusted for Standard requirements
Google Gemini 3.1 Pro Preview
one-step LLM adjusted for Standard requirements
OpenAI GPT-5.6 Sol
two-step
OpenAI GPT-5.6 Sol
two-step
Google Gemini 3.1 Pro Preview
two-step
OpenAI GPT-5.6 Sol
two-step
OpenAI GPT-5.6 Sol
two-step
Google Gemini 3.1 Pro Preview
two-step
Optimize for selects one of the solutions that meet the level: the one with the fewest total errors, the lowest cost or the shortest time. Where no solution meets the level, the card shows the available solution with the fewest total errors and what it misses. Total errors count every requirement with countable errors.
| Quality level | Tier | Solutions | Simplest solution | Model | Total errors per 1,000 words | Cost per million characters | Seconds per 1,000 words |
|---|---|---|---|---|---|---|---|
| Basic | Tier 1: simple solution + backup | ★★ | one-step | Google Gemini 3.1 Pro Preview | 17.1 | $6.39 | 84 |
| Standard | Tier 2: two or more solutions, none simple | ☆☆ | two-step | OpenAI GPT-5.6 Sol | 16.5 | $27.98 | 220 |
| Advanced | Tier 2: two or more solutions, none simple | ☆☆ | two-step | OpenAI GPT-5.6 Sol | 16.5 | $27.98 | 220 |
One solution per level: the simplest that meets the level, the cheapest of them where several are equally simple. ★ a one-step solution, ☆ a solution with more steps or the gender-neutrality agent.
The step up from Basic to Standard removes both one-step solutions that meet every Basic requirement in Lao. Translations from the one on Google Gemini 3.1 Pro Preview have 97.9% of the sentences in the formal register, and those from the one on OpenAI GPT-5.6 Sol 99.6% of the tags right. The two-step solutions (translation + post-edit) on both models meet every Standard and Advanced requirement. Optimizing for cost selects Google Gemini 3.1 Pro Preview under the Basic requirements and OpenAI GPT-5.6 Sol under the others.
| Model | US dollars per million characters | Weighted errors per 1,000 words | Group |
|---|---|---|---|
| DeepSeek V3.2 | 2.97 | 53.7 | Misses a Standard requirement |
| Google Gemini 3.5 Flash | 4.79 | 101.8 | Misses a Standard requirement |
| Google Gemini 3.1 Pro Preview | 6.39 | 16.8 | Misses a Standard requirement |
| z.ai GLM-5 | 7.1 | 83.4 | Misses a Standard requirement |
| Microsoft Translator | 10 | 234.7 | Misses a Standard requirement |
| Anthropic Claude Sonnet 4.6 | 15.96 | 50.8 | Misses a Standard requirement |
| Google Translation LLM | 20.9 | 85.7 | Misses a Standard requirement |
| Translated Lara 2 | 24.99 | 83.4 | Misses a Standard requirement |
| OpenAI GPT-5.6 Sol | 27.98 | 16.8 | Misses a Standard requirement |
| OpenAI GPT-5.5 | 41.14 | 22.1 | Misses a Standard requirement |
In Lao, the one-step solution on Google Gemini 3.1 Pro Preview costs $6.39 per million characters and has 16.8 weighted errors per 1,000 words. None of the dearer ones have fewer weighted errors per 1,000 words, and the one on OpenAI GPT-5.6 Sol at $27.98 per million characters has as many. The dearest, on OpenAI GPT-5.5 at $41.14 per million characters, has 22.1 weighted errors per 1,000 words. The cheaper ones with Intento LQA results, on DeepSeek V3.2 and Google Gemini 3.5 Flash, have 53.7 and 101.8 weighted errors per 1,000 words.
| Requirement | Median errors per 1,000 words |
|---|---|
| General translation quality | 68.6 |
| Tag handling | 1 |
| Tone of voice | 0.3 |
| Gender neutrality | 0 |
The requirements with countable errors. Every solution with its result on each requirement: the explorer.
Among the one-step solutions that do not meet general translation quality in Lao, the one on OpenAI GPT-5.5 has the fewest weighted errors, 22.1 per 1,000 words, with no major or critical error. The two-step solution (translation + post-edit) on the same model has 23.1 weighted errors per 1,000 words. The one-step solutions on all seven shortlisted LLMs meet formatting. None of the solutions on the open-weight Google Gemma 4, Microsoft Translator and the two LLM-based MT engines, Translated Lara 2 and Google Translation LLM, meet it.
| Model | Weighted errors per 1,000 words of the shortlisted one-step solutions in Lao | Class |
|---|---|---|
| Microsoft Translator | 234.7 | MT engines |
| Google Gemini 3.5 Flash | 101.8 | LLMs |
| Google Translation LLM | 85.7 | LLM-based MT engines |
| z.ai GLM-5 | 83.4 | LLMs |
| Translated Lara 2 | 83.4 | LLM-based MT engines |
| DeepSeek V3.2 | 53.7 | LLMs |
| Anthropic Claude Sonnet 4.6 | 50.8 | LLMs |
| OpenAI GPT-5.5 | 22.1 | LLMs |
| Google Gemini 3.1 Pro Preview | 16.8 | LLMs |
| OpenAI GPT-5.6 Sol | 16.8 | LLMs |
No translation into Lao: AI21 Jamba Large 1.7 (garbage or near-empty, wrong language, wrong script), Swiss AI Apertus-8B-2509 (cut off part-way, wrong language, wrong script, far too short), MiniMax M2.5 (garbage or near-empty, wrong language, wrong script), OpenAI gpt-oss-120b (garbage or near-empty).
The shortlisted LLMs that meet our language-support criteria in Lao are Google Gemini 3.1 Pro Preview, OpenAI GPT-5.6 Sol, OpenAI GPT-5.5 and DeepSeek V3.2. Of the other three, translations from Google Gemini 3.5 Flash and Anthropic Claude Sonnet 4.6 have 18 and 4 critical errors, and those from z.ai GLM-5 83.4 weighted errors per 1,000 words. Translations from the MT engine Microsoft Translator have 234.7 weighted errors per 1,000 words. The one-step solution on the open-weight Google Gemma 4 has no conclusive Intento LQA result.
| Model | COMET-Kiwi | Rank on COMET-Kiwi | MetricX | Rank on MetricX | Shortlisted | Shortlist place |
|---|---|---|---|---|---|---|
| Mistral Large 3 | 0.762 | 11 | 1.66 | 10 | No | Not shortlisted |
| Anthropic Claude Fable 5 | 0.761 | 12 | 1.89 | 14 | No | Not shortlisted |
| MoonshotAI Kimi K2.5 | 0.760 | 13 | 1.73 | 12 | No | Not shortlisted |
| Anthropic Claude Sonnet 5 | 0.755 | 14 | 1.81 | 13 | No | Not shortlisted |
| OpenAI GPT-4.1 | 0.753 | 15 | 1.92 | 15 | No | Not shortlisted |
| xAI Grok 4.20 (non-reasoning) | 0.751 | 16 | 1.94 | 16 | No | Not shortlisted |
| Anthropic Claude Opus 4.8 | 0.746 | 17 | 2.05 | 19 | No | Not shortlisted |
| Google Cloud Advanced Translation API | 0.738 | 19 | 2.47 | 23 | No | Not shortlisted |
| Alibaba Qwen3-235B-A22B | 0.730 | 20 | 2.36 | 22 | No | Not shortlisted |
| NiuTrans | 0.730 | 21 | 1.98 | 18 | No | Not shortlisted |
| Anthropic Claude Haiku 4.5 | 0.727 | 22 | 2.10 | 20 | No | Not shortlisted |
| ModernMT Enterprise Edition | 0.721 | 23 | 2.13 | 21 | No | Not shortlisted |
| Cohere Tiny Aya | 0.668 | 24 | 4.17 | 26 | No | Not shortlisted |
| Mistral Medium 3.1 | 0.651 | 25 | 3.76 | 24 | No | Not shortlisted |
| Yandex Cloud Translate v2 | 0.649 | 26 | 3.99 | 25 | No | Not shortlisted |
| Alibaba Machine Translation API | 0.276 | 27 | 13.17 | 27 | No | Not shortlisted |
| Cohere Command A Translate | 0.258 | 28 | 16.35 | 28 | No | Not shortlisted |
| Google Gemini 3.1 Pro Preview | 0.783 | 1 | 1.34 | 2 | Yes | Top 8 by both metrics |
| Google Gemini 3.5 Flash | 0.778 | 2 | 1.29 | 1 | Yes | Top 8 by both metrics |
| OpenAI GPT-5.6 Sol | 0.777 | 3 | 1.39 | 3 | Yes | Top 8 by both metrics |
| Google Gemma 4 | 0.770 | 4 | 1.45 | 4 | Yes | Top 8 by both metrics |
| z.ai GLM-5 | 0.770 | 5 | 1.53 | 6 | Yes | Top 8 by both metrics |
| DeepSeek V3.2 | 0.770 | 6 | 1.45 | 5 | Yes | Top 8 by both metrics |
| Anthropic Claude Sonnet 4.6 | 0.768 | 7 | 1.62 | 9 | Yes | Top 8 by COMET-Kiwi |
| OpenAI GPT-5.5 | 0.768 | 8 | 1.57 | 7 | Yes | Top 8 by both metrics |
| Translated Lara 2 | 0.767 | 9 | 1.58 | 8 | Yes | Top 8 by MetricX |
| Google Translation LLM | 0.764 | 10 | 1.72 | 11 | Yes | Real-time place |
| Microsoft Translator | 0.742 | 18 | 1.95 | 17 | Yes | Real-time place |
| Model | Solution complexity | COMET-Kiwi rank | COMET-Kiwi | MetricX rank | MetricX | Shortlist |
|---|---|---|---|---|---|---|
| Google Gemini 3.1 Pro Preview | Adjusted LLM | 1 | 0.783 | 2 | 1.34 | Top 8 by both metrics |
| Google Gemini 3.5 Flash | Adjusted LLM | 2 | 0.778 | 1 | 1.29 | Top 8 by both metrics |
| OpenAI GPT-5.6 Sol | Adjusted LLM | 3 | 0.777 | 3 | 1.39 | Top 8 by both metrics |
| Google Gemma 4 | Adjusted LLM | 4 | 0.770 | 4 | 1.45 | Top 8 by both metrics |
| z.ai GLM-5 | Adjusted LLM | 5 | 0.770 | 6 | 1.53 | Top 8 by both metrics |
| DeepSeek V3.2 | Adjusted LLM | 6 | 0.770 | 5 | 1.45 | Top 8 by both metrics |
| Anthropic Claude Sonnet 4.6 | Adjusted LLM | 7 | 0.768 | 9 | 1.62 | Top 8 by COMET-Kiwi |
| OpenAI GPT-5.5 | Adjusted LLM | 8 | 0.768 | 7 | 1.57 | Top 8 by both metrics |
| Translated Lara 2 | Adjusted NMT | 9 | 0.767 | 8 | 1.58 | Top 8 by MetricX |
| Google Translation LLM | Adjusted NMT | 10 | 0.764 | 11 | 1.72 | Real-time place |
| Mistral Large 3 | Adjusted LLM | 11 | 0.762 | 10 | 1.66 | Not shortlisted |
| Anthropic Claude Fable 5 | Adjusted LLM | 12 | 0.761 | 14 | 1.89 | Not shortlisted |
| MoonshotAI Kimi K2.5 | Adjusted LLM | 13 | 0.760 | 12 | 1.73 | Not shortlisted |
| Anthropic Claude Sonnet 5 | Adjusted LLM | 14 | 0.755 | 13 | 1.81 | Not shortlisted |
| OpenAI GPT-4.1 | Adjusted LLM | 15 | 0.753 | 15 | 1.92 | Not shortlisted |
| xAI Grok 4.20 (non-reasoning) | Adjusted LLM | 16 | 0.751 | 16 | 1.94 | Not shortlisted |
| Anthropic Claude Opus 4.8 | Adjusted LLM | 17 | 0.746 | 19 | 2.05 | Not shortlisted |
| Microsoft Translator | Off-the-shelf NMT | 18 | 0.742 | 17 | 1.95 | Real-time place |
| Google Cloud Advanced Translation API | Off-the-shelf NMT | 19 | 0.738 | 23 | 2.47 | Not shortlisted |
| Alibaba Qwen3-235B-A22B | Adjusted LLM | 20 | 0.730 | 22 | 2.36 | Not shortlisted |
| NiuTrans | Off-the-shelf NMT | 21 | 0.730 | 18 | 1.98 | Not shortlisted |
| Anthropic Claude Haiku 4.5 | Adjusted LLM | 22 | 0.727 | 20 | 2.10 | Not shortlisted |
| ModernMT Enterprise Edition | Off-the-shelf NMT | 23 | 0.721 | 21 | 2.13 | Not shortlisted |
| Cohere Tiny Aya | Off-the-shelf LLM | 24 | 0.668 | 26 | 4.17 | Not shortlisted |
| Mistral Medium 3.1 | Adjusted LLM | 25 | 0.651 | 24 | 3.76 | Not shortlisted |
| Yandex Cloud Translate v2 | Off-the-shelf NMT | 26 | 0.649 | 25 | 3.99 | Not shortlisted |
| Alibaba Machine Translation API | Off-the-shelf NMT | 27 | 0.276 | 27 | 13.17 | Not shortlisted |
| Cohere Command A Translate | Adjusted LLM | 28 | 0.258 | 28 | 16.35 | Not shortlisted |
| OpenAI GPT-5.6 Luna | Adjusted LLM | — | 0.772 | — | 1.45 | Same model family as OpenAI GPT-5.6 Sol |
| OpenAI GPT-5.6 Terra | Adjusted LLM | — | 0.770 | — | 1.47 | Same model family as OpenAI GPT-5.6 Sol |
| xAI Grok 4.20 (reasoning) | Adjusted LLM | — | 0.744 | — | 2.14 | Same model family as xAI Grok 4.20 (non-reasoning) |
Every model we got a translation from, as the shortlist compared it: off the shelf, without context in the request, or adjusted, with it. A model that did not make the shortlist is shown adjusted, unless off the shelf it scores higher by 0.03 or more on COMET-Kiwi. Ranks among 28 models, with each model family counted once, and the top 8 of each metric in bold. Higher is better on COMET-Kiwi, lower on MetricX. The scores rank models within this language and are not a requirement result.
MetricX ranks Google Gemini 3.5 Flash first in Lao and COMET-Kiwi second, yet by Intento LQA its one-step solution has 101.8 weighted errors per 1,000 words and 18 critical errors. The one-step solutions with the fewest weighted errors, on Google Gemini 3.1 Pro Preview and OpenAI GPT-5.6 Sol, rank in the top 3 on both metrics. COMET-Kiwi alone put Anthropic Claude Sonnet 4.6 on the shortlist, and MetricX alone Translated Lara 2. Nearest the cut is Mistral Large 3, the first model of the tail on both COMET-Kiwi and MetricX.
| Quality level | Cheapest solution that meets every requirement of the level | Cost per million characters |
|---|---|---|
| Basic | Google Gemini 3.1 Pro Preview, one-step LLM adjusted for Standard requirements | $6.39 |
| Standard | OpenAI GPT-5.6 Sol, two-step | $27.98 |
| Advanced | OpenAI GPT-5.6 Sol, two-step | $27.98 |
| Model | US dollars per million characters | Class |
|---|---|---|
| Google Gemma 4 | $0.19 | Open-weight LLMs |
| DeepSeek V3.2 | $2.97 | LLMs |
| Google Gemini 3.5 Flash | $4.79 | LLMs |
| Google Gemini 3.1 Pro Preview | $6.39 | LLMs |
| z.ai GLM-5 | $7.1 | LLMs |
| Microsoft Translator | $10 | MT engines |
| Anthropic Claude Sonnet 4.6 | $15.96 | LLMs |
| Google Translation LLM | $20.9 | LLM-based MT engines |
| Translated Lara 2 | $24.99 | LLM-based MT engines |
| OpenAI GPT-5.6 Sol | $27.98 | LLMs |
| OpenAI GPT-5.5 | $41.14 | LLMs |
| Model | Tokens per character | Class |
|---|---|---|
| Mistral Large 3 | 1.918 | LLMs |
| Mistral Medium 3.1 | 1.901 | LLMs |
| Cohere Command A Translate | 1.736 | LLMs |
| OpenAI GPT-4.1 | 1.579 | LLMs |
| z.ai GLM-5 | 1.563 | LLMs |
| OpenAI GPT-5.6 Sol | 1.353 | LLMs |
| OpenAI GPT-5.6 Terra | 1.333 | LLMs |
| OpenAI GPT-5.5 | 1.328 | LLMs |
| OpenAI GPT-5.6 Luna | 1.323 | LLMs |
| DeepSeek V3.2 | 1.315 | LLMs |
| Anthropic Claude Opus 4.8 | 1.072 | LLMs |
| Anthropic Claude Fable 5 | 1.07 | LLMs |
| xAI Grok 4.20 (non-reasoning) | 1.064 | LLMs |
| MoonshotAI Kimi K2.5 | 1.052 | LLMs |
| xAI Grok 4.20 (reasoning) | 1.048 | LLMs |
| Anthropic Claude Sonnet 4.6 | 1.041 | LLMs |
| Anthropic Claude Haiku 4.5 | 1.031 | LLMs |
| Alibaba Qwen3-235B-A22B | 0.975 | Open-weight LLMs |
| Anthropic Claude Sonnet 5 | 0.972 | LLMs |
| Google Gemma 4 | 0.469 | Open-weight LLMs |
| Google Gemini 3.1 Pro Preview | 0.467 | LLMs |
| Google Gemini 3.5 Flash | 0.465 | LLMs |
| Cohere Tiny Aya | 0.393 | LLMs |
Translations into Lao take 0.467 tokens per character on Google Gemini 3.1 Pro Preview and 1.353 on OpenAI GPT-5.6 Sol. One-step solutions on the two cost $6.39 and $27.98 per million characters. Even so, the one on DeepSeek V3.2 costs $2.97 per million characters, at 1.315 tokens per character. The post-edit step raises the cost on Google Gemini 3.1 Pro Preview to $28.72 per million characters. The cheapest solution goes from $6.39 under the Basic requirements to $27.98 per million characters under the Standard and Advanced ones.
| Quality level | Fastest solution that meets every requirement of the level | Seconds per 1,000 words |
|---|---|---|
| Basic | Google Gemini 3.1 Pro Preview, one-step LLM adjusted for Standard requirements | 84 |
| Standard | Google Gemini 3.1 Pro Preview, two-step | 163 |
| Advanced | Google Gemini 3.1 Pro Preview, two-step | 163 |
| Model | Seconds per 1,000 words | Class |
|---|---|---|
| Google Cloud Advanced Translation API | 0.1 | MT engines |
| Microsoft Translator | 0.2 | MT engines |
| ModernMT Enterprise Edition | 0.4 | MT engines |
| Google Translation LLM | 0.7 | LLM-based MT engines |
| Yandex Cloud Translate v2 | 4.3 | MT engines |
| Translated Lara 2 | 5.5 | LLM-based MT engines |
| Cohere Tiny Aya | 22 | LLMs |
| Google Gemini 3.5 Flash | 23 | LLMs |
| Alibaba Machine Translation API | 30 | MT engines |
| OpenAI GPT-5.6 Luna | 50 | LLMs |
| NiuTrans | 65 | MT engines |
| OpenAI GPT-5.6 Terra | 77 | LLMs |
| DeepSeek V3.2 | 78 | LLMs |
| Google Gemini 3.1 Pro Preview | 84 | LLMs |
| xAI Grok 4.20 (non-reasoning) | 91 | LLMs |
| Anthropic Claude Haiku 4.5 | 100 | LLMs |
| Cohere Command A Translate | 110 | LLMs |
| OpenAI GPT-4.1 | 116 | LLMs |
| Mistral Medium 3.1 | 171 | LLMs |
| Anthropic Claude Sonnet 4.6 | 176 | LLMs |
| xAI Grok 4.20 (reasoning) | 201 | LLMs |
| Google Gemma 4 | 202 | Open-weight LLMs |
| OpenAI GPT-5.6 Sol | 220 | LLMs |
| OpenAI GPT-5.5 | 266 | LLMs |
| Anthropic Claude Sonnet 5 | 281 | LLMs |
| Mistral Large 3 | 283 | LLMs |
| Alibaba Qwen3-235B-A22B | 341 | Open-weight LLMs |
| Anthropic Claude Fable 5 | 346 | LLMs |
| Anthropic Claude Opus 4.8 | 377 | LLMs |
| MoonshotAI Kimi K2.5 | 484 | LLMs |
| z.ai GLM-5 | 3,897 | 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 Lao.
All five real-time models in Lao are MT engines, two of them LLM-based. Three of them, Microsoft Translator, Google Translation LLM and Translated Lara 2, are on the shortlist. Their solutions miss general translation quality, tag handling and formatting, and those on Google Translation LLM and Translated Lara 2 meet tone of voice. On Google Gemini 3.1 Pro Preview, the fastest solutions that meet every Basic and every Standard requirement take 84 and 163 seconds per 1,000 words, in one step and in two steps (translation + post-edit).
| The two best solutions | Solution | Sentences about people that are gender-neutral | Meets every Advanced requirement | Cost per million characters | Seconds per 1,000 words |
|---|---|---|---|---|---|
| OpenAI GPT-5.6 Sol | two-step | 100% | ✓ | $27.98 | 220 |
| Google Gemini 3.1 Pro Preview | two-step | 100% | ✓ | $28.72 | 162 |
For Lao we check the formal register, with respectful vocabulary, and gender-neutral wording with no default gender. Lao has no grammatical gender for persons, so a translation is gender-neutral by default, gendered terms aside. The two best solutions, both two-step (translation + post-edit), have 100% of the sentences about people gender-neutral without the gender-neutrality agent. In one step, translations from the shortlisted LLMs have 97.9% to 100% of the sentences in the formal register, and those from Microsoft Translator 83.0%.
The three LQA judges agree on a median 92.7% of the sentences across the 11 shortlisted solutions in Lao, so the Intento LQA error counts on this page carry weight. The quality-estimation pair of automatic metrics has an agreement of 0.69, with a range of 0.68 to 0.70 above its threshold of 0.53. The similarity pair has 0.41, and even the top of its range, 0.42, stays under its threshold of 0.46.
Leave feedback