About
Custom.MT is a specialized generative AI and machine translation platform built for localization teams and language service providers. It centralizes the management of both stock and custom translation models, enabling organizations to benchmark engines—including DeepL, Google Translate, GPT-4, and proprietary models—side by side using the Best Engine tool in the Custom.MT Console with BLEU, WER, and COMET scoring. The platform offers MT model fine-tuning so teams can train models on domain-specific terminology and style guides, reducing post-editing effort and driving down localization costs. For organizations handling sensitive content, Custom.MT supports fully on-premise deployment, keeping data off public cloud infrastructure. Additional capabilities include Translation Memory (TMX) Cleaning to keep linguistic assets accurate and ready for training, Machine Translation Automatic Post-Editing (MTAPE) that can cut linguistic audit costs by up to 70%, Machine Translation Quality Estimation, and language dataset acquisition across specialized domains like medical, legal, patent, and eCommerce. Custom.MT integrates natively with leading TMS and CAT tools—Trados, Smartling, memoQ, XTM Cloud, and Smartcat—as well as Shopware for eCommerce translation. An API is available for custom integrations. The platform is suited for enterprise localization departments, translation agencies, and language technology teams looking to scale quality-controlled multilingual content production.
Key Features
- Multi-Engine MT Evaluation: Compare DeepL, Google Translate, GPT-4, and custom models using BLEU, WER, and COMET metrics in the Custom.MT Console to select the best engine for each language pair or domain.
- MT Model Fine-Tuning: Train and fine-tune machine translation models on domain-specific terminology and style guides to improve accuracy and reduce post-editing distance.
- On-Premise Deployment: Deploy machine translation on your own hardware infrastructure to process confidential or regulated content without exposing it to public cloud services.
- Automatic MT Post-Editing: Replace manual linguistic reviews with automated post-editing workflows that can reduce linguistic audit costs by up to 70%.
- TMS & CAT Tool Integrations: Native integrations with Trados, Smartling, memoQ, XTM Cloud, and Smartcat, plus a Shopware plugin and open API for custom workflows.
Use Cases
- Localization teams evaluating and selecting the best-performing MT or LLM engine for specific language pairs or content domains using objective quality metrics.
- Enterprises fine-tuning machine translation models with proprietary terminology and brand style guides to achieve higher translation consistency and lower post-editing costs.
- Regulated-industry organizations (legal, medical, financial) requiring fully on-premise MT deployment to comply with data privacy and confidentiality policies.
- Translation agencies automating the post-editing review stage with MTAPE to significantly reduce per-word linguistic audit costs at scale.
- Language technology teams acquiring domain-specific training datasets (medical, legal, patent, eCommerce) and cleaning translation memories (TMX) to improve model training data quality.
Pros
- Objective Engine Benchmarking: Side-by-side comparison of multiple MT and LLM engines with industry-standard metrics (BLEU, WER, COMET) removes guesswork from engine selection.
- Broad TMS Ecosystem Integration: Out-of-the-box connectors for the most widely used translation management and CAT tools allow teams to plug Custom.MT into existing localization pipelines.
- On-Premise Option for Sensitive Content: Organizations in regulated industries can run MT models entirely on their own infrastructure, satisfying data privacy and compliance requirements.
- End-to-End Localization Toolset: Covers the full MT lifecycle—data acquisition, model training, quality evaluation, post-editing, and TMX cleaning—within a single platform.
Cons
- Steep Learning Curve for MT Metrics: Getting value from BLEU, WER, and COMET scores requires familiarity with MT evaluation concepts, which may be challenging for non-technical localization managers.
- Enterprise-Oriented Pricing: The platform is primarily designed for enterprise localization teams; smaller agencies or freelance translators may find the feature set and pricing model disproportionate to their needs.
- Limited Pricing Transparency: Enterprise plans require booking a discovery call, making it difficult to estimate costs without engaging the sales team.
Frequently Asked Questions
Which machine translation engines does Custom.MT support?
Custom.MT supports a wide range of MT and LLM engines, including DeepL, Google Translate, GPT-4, and custom or fine-tuned models. Teams can compare all engines side by side in the Custom.MT Console.
Can Custom.MT be deployed on-premise?
Yes. Custom.MT offers an on-premise machine translation deployment option, allowing organizations to run MT models on their own hardware for content that must not leave their internal infrastructure.
What quality metrics does Custom.MT use to evaluate MT engines?
Custom.MT uses BLEU (Bilingual Evaluation Understudy), WER (Word Error Rate), and COMET scores, along with human evaluation workflows, to assess and compare model performance across language pairs and domains.
Which TMS and CAT tools does Custom.MT integrate with?
Custom.MT has native integrations for Trados, Smartling, memoQ, XTM Cloud, and Smartcat, as well as a Shopware Translation Plugin for eCommerce. An open API is also available for custom integrations.
Is there a free trial available?
Yes, Custom.MT offers a free trial through the Custom.MT Console. Users can sign up and start translating to explore the platform's features before committing to a paid plan.
