# Best machine translation APIs for AI agents > Azure Translator (BB), DeepL API (BB) and Lokalise (BB) lead the 12 ranked machine translation APIs. Picks by need, strengths, weaknesses and prices from the Anchor benchmark. - Canonical: https://www.anchorterminal.com/best/translation/ - Markdown: https://www.anchorterminal.com/best/translation/index.md (~5,850 tokens) - Slim: https://www.anchorterminal.com/best/translation/index.min.md (~1,530 tokens, same facts, less prose, for token-sensitive contexts) - JSON: https://www.anchorterminal.com/best/translation/index.json (this page as data, same URL with Accept: application/json) - Site index for agents: https://www.anchorterminal.com/llms.txt (full text: https://www.anchorterminal.com/llms-full.txt) - API: https://www.anchorterminal.com/api/v1/index.json - Updated: 2026-10-08 The 10 highest-scoring of 12 machine translation APIs on the Anchor benchmark, with a pick for each need and where each one falls short. Scores come from public evidence, re-checked as vendors change. - Ranked: 12 · agent-ready (BB or better): 3 · accept x402: 0 · hosted endpoints: 12 - Full ranked table: https://www.anchorterminal.com/categories/translation.md - Head-to-head comparisons: https://www.anchorterminal.com/compare/translation/index.md (66) - Methodology: https://www.anchorterminal.com/benchmark/index.md ## The shortlist | # | Tool | Grade | Score | Best for | Price | Where | | --- | --- | --- | --- | --- | --- | --- | | 1 | [Azure Translator](https://www.anchorterminal.com/tools/azure-translator.md) | BB | 72 | Azure shops that want cheap NMT at volume, Entra ID auth and a choice of an LLM per request for tone and gender. | $10 / mo | hosted | | 2 | [DeepL API](https://www.anchorterminal.com/tools/deepl-api.md) | BB | 72 | Agents that need terminology control, formality and document layout kept, in European and Asian languages, and teams that want an MCP server today. | Freemium | hosted and local | | 3 | [Lokalise](https://www.anchorterminal.com/tools/lokalise.md) | BB | 71.3 | Agents that manage localisation work, such as adding keys, uploading and exporting string files, keeping a glossary and starting AI or machine translation tasks for a team that already uses Lokalise. | $149 / mo | hosted | | 4 | [Crowdin](https://www.anchorterminal.com/tools/crowdin.md) | B | 69.6 | Teams that keep product strings and files in Crowdin and want an agent to upload sources, check progress, apply pre-translation, manage glossaries and pull builds. | $50 / mo | hosted | | 5 | [Amazon Translate](https://www.anchorterminal.com/tools/amazon-translate.md) | B | 69.5 | Operators already on AWS who want translation inside IAM and CloudTrail, with S3 batch jobs and terminology files. | Pay per use | hosted | | 6 | [Google Cloud Translation](https://www.anchorterminal.com/tools/google-cloud-translation.md) | B | 68.1 | Agents on Google Cloud that want high quotas, clear data terms and a choice of NMT, LLM or adaptive translation with glossaries. | $80 / 1k pages | hosted | | 7 | [Phrase](https://www.anchorterminal.com/tools/phrase.md) | B | 64.2 | Teams that keep translation projects, term bases and translation memories in Phrase and want an agent to run them, or to translate through engines chosen per language pair. | $27 / mo | hosted | | 8 | [Lingo.dev](https://www.anchorterminal.com/tools/lingo-dev.md) | B | 62.3 | Product teams that translate interface strings and content into many locales and want glossary, tone and rule control held in one place, driven from code, CI or a coding assistant. | $99 / mo | hosted | | 9 | [Lara Translate API](https://www.anchorterminal.com/tools/lara-translate.md) | C | 61.3 | Agents that localise for one brand and need glossaries, translation memories and style instructions on every call, especially from an MCP client. | Freemium | hosted and local | | 10 | [LibreTranslate](https://www.anchorterminal.com/tools/libretranslate.md) | D | 49.8 | Operators who must keep text on their own servers or want a fixed cost, and for light hosted use. | $29 / mo | hosted | ## Picks by need - Highest score overall: [Azure Translator](https://www.anchorterminal.com/tools/azure-translator.md), BB, 72/100 on the benchmark. Also [DeepL API](https://www.anchorterminal.com/tools/deepl-api.md), BB, 72/100. - Reliability: [Amazon Translate](https://www.anchorterminal.com/tools/amazon-translate.md), 90/100 on reliability, against 83 for the overall leader. - Schema & documentation: [DeepL API](https://www.anchorterminal.com/tools/deepl-api.md), 95/100 on schema & documentation, against 83 for the overall leader. - Agent ergonomics: [DeepL API](https://www.anchorterminal.com/tools/deepl-api.md), 83/100 on agent ergonomics, against 82 for the overall leader. - Security & auth: [DeepL API](https://www.anchorterminal.com/tools/deepl-api.md), 82/100 on security & auth, against 75 for the overall leader. - Maintenance & community: [DeepL API](https://www.anchorterminal.com/tools/deepl-api.md), 87/100 on maintenance & community, against 70 for the overall leader. - Transparency & trust: [Lokalise](https://www.anchorterminal.com/tools/lokalise.md), 81/100 on transparency & trust, against 78 for the overall leader. - A hosted MCP endpoint: [DeepL API](https://www.anchorterminal.com/tools/deepl-api.md), remote MCP server, nothing to install. - Self-hosting under an open licence: [LibreTranslate](https://www.anchorterminal.com/tools/libretranslate.md), self-hosted, AGPL-3 licence. ## How to choose - Quality in your language pairs: Test the language pairs the agent will use with your own sample texts, since quality varies between pairs and a benchmark for one pair does not carry over. - Glossary terms in the output: Check that glossary terms survive into the output, because an agent cannot easily spot a product name the translation has changed silently. - Formatting kept in documents: Test a formatted document to see whether tables, headings and markup come back intact, since an agent that re-parses broken output has to rebuild the layout. - Retention and training use: Read how long submitted text and files are kept and whether they are used for training, because agent prompts may carry customer data. - How the benchmark tests this category: The same texts in six language pairs, one with a glossary and one as a formatted document. We check the glossary terms, the formatting, latency, what the provider keeps and the cost per million characters. ## Each one in detail ### 1. Azure Translator, BB 72/100 Microsoft's translation API, now part of Foundry Tools. - Verdict: $10 per million characters pay as you go, 2 million free a month on F0, commitment overage down to $5.50. The 2026-06-06 schema breaks v3.0 clients and drops BreakSentence and dictionary lookups from the spec. - Choose it for: Azure shops that want cheap NMT at volume, Entra ID auth and a choice of an LLM per request for tone and gender. - Strength: $10 per million characters pay as you go, 2 million free a month on F0, commitment overage down to $5.50 - Strength: Text isn't stored and documents are hard-deleted after processing - Strength: OpenAPI documents for 3.0 and 2026-06-06 in Microsoft's public spec repository - Weakness: The 2026-06-06 schema breaks v3.0 clients and drops BreakSentence and dictionary lookups from the spec - Weakness: LLM translation needs a Foundry deployment and bills tokens on the Azure OpenAI meter - Weakness: Tone and gender controls only on the LLM option, which takes 50 texts of 5,000 characters a call - Price: $10 / mo · Auth: OAuth or key · x402: no · Where: hosted - Full assessment: https://www.anchorterminal.com/tools/azure-translator.md ### 2. DeepL API, BB 72/100 Translation API for text and documents in over 100 languages, with glossaries, formality control, style rules, custom instructions, a writing endpoint and a voice API. - Verdict: Up to 5 glossaries, five formality settings, style rules and up to 10 custom instructions on the translate call. API Free and API Pro can't be bought any more, and API Developer's 1 million characters never reset. - Choose it for: Agents that need terminology control, formality and document layout kept, in European and Asian languages, and teams that want an MCP server today. - Strength: Up to 5 glossaries, five formality settings, style rules and up to 10 custom instructions on the translate call - Strength: OpenAPI document synced daily to GitHub, llms.txt and 17 dated changelog entries since 1 July 2026 - Strength: API Key Permissions scope a key to chosen endpoints, and keys travel only in a header - Weakness: API Free and API Pro can't be bought any more, and API Developer's 1 million characters never reset - Weakness: Per-character rates only appear on a browser-rendered pricing page - Weakness: No published request rate and no SLA found for self-serve plans - Price: Freemium · Auth: OAuth or key · x402: no · Where: hosted and local - Full assessment: https://www.anchorterminal.com/tools/deepl-api.md - Against #1: https://www.anchorterminal.com/compare/azure-translator-vs-deepl-api.md ### 3. Lokalise, BB 71.3/100 Lokalise is a hosted translation management system for software strings, documents and marketing content. Agents reach it through a REST API, the `lokalise2` CLI, SDKs in four languages and a hosted MCP server in beta. - Verdict: The REST API has public OpenAPI files, Markdown docs, published rate limits, read-only tokens, scoped OAuth and an audit log endpoint. Access needs a paid plan after a 14-day trial, a person signs up in a browser, and the Master Service Agreement forbids benchmarking. No idempotency keys or deprecation policy were found in the reviewed documentation. - Choose it for: Agents that manage localisation work, such as adding keys, uploading and exporting string files, keeping a glossary and starting AI or machine translation tasks for a team that already uses Lokalise. - Strength: Three public OpenAPI 3.0.3 files, llms.txt and a Markdown twin of every reference page - Strength: Rate limits published as 6 requests a second per token and per IP, with 10 concurrent requests per project - Strength: API tokens are read-only or read/write, and OAuth 2 carries read and write scopes per resource - Weakness: The Master Service Agreement forbids using the service for benchmarking or competitive analysis - Weakness: No free plan. Explorer is $149 a month billed yearly after a 14-day trial without a card - Weakness: No idempotency keys in the OpenAPI files, though the errors page mentions an idempotent key under 409 - Price: $149 / mo · Auth: OAuth or key · x402: no · Where: hosted - Full assessment: https://www.anchorterminal.com/tools/lokalise.md - Against #1: https://www.anchorterminal.com/compare/azure-translator-vs-lokalise.md ### 4. Crowdin, B 69.6/100 Crowdin is a hosted localisation platform that stores source files and strings, translation memories and glossaries, and runs machine and AI translation over them. Agents reach it through a REST API, a CLI and a hosted MCP server. - Verdict: The REST API has a public OpenAPI document with 374 operations, and tokens can be limited by scope, access level and project. Translation runs inside Crowdin projects and connected engines, so it is not a stateless per-character translation API. No Retry-After guidance, idempotency keys or SLA were found in the reviewed documentation. - Choose it for: Teams that keep product strings and files in Crowdin and want an agent to upload sources, check progress, apply pre-translation, manage glossaries and pull builds. - Strength: Public OpenAPI 3.0 document with 374 operations on 218 paths, plus `llms.txt` with one text file per route - Strength: Personal access tokens take an expiry date, per-scope read-only or read-and-write access, and a granular option limiting them to chosen projects - Strength: The MCP V2 endpoint filters tools by header, with `crowdin-read-only`, `crowdin-tool-sets`, `crowdin-project-type` and `crowdin-tool-filter` - Weakness: The terms forbid accessing the service to monitor availability, performance or functionality, or for benchmarking, without Crowdin's prior written consent - Weakness: A 429 is returned above 20 simultaneous requests an account, with no Retry-After header or backoff guidance in the reference - Weakness: No idempotency keys in the OpenAPI document, and no SLA found in the terms or the help centre - Price: $50 / mo · Auth: OAuth or key · x402: no · Where: hosted - Full assessment: https://www.anchorterminal.com/tools/crowdin.md - Against #1: https://www.anchorterminal.com/compare/azure-translator-vs-crowdin.md ### 5. Amazon Translate, B 69.5/100 AWS's neural translation API across 75 languages. - Verdict: IAM policies scope access per action and resource, with a `TranslateReadOnly` managed policy, and CloudTrail logs each call. AWS may store and use text to improve Amazon AI services unless the organisation sets an opt-out policy. - Choose it for: Operators already on AWS who want translation inside IAM and CloudTrail, with S3 batch jobs and terminology files. - Strength: IAM policies scope access per action and resource, with a `TranslateReadOnly` managed policy, and CloudTrail logs each call - Strength: Formality, profanity masking and brevity on `TranslateText`, reported back in `AppliedSettings` - Strength: S3 batch jobs up to 5 GB and 1 million documents, with a `ClientToken` for idempotency - Weakness: AWS may store and use text to improve Amazon AI services unless the organisation sets an opt-out policy - Weakness: No new capability in the document history since 2023-10-31 - Weakness: 10,000 bytes a `TranslateText` call and one string per call - Price: Pay per use · Auth: API key · x402: no · Where: hosted - Full assessment: https://www.anchorterminal.com/tools/amazon-translate.md - Against #1: https://www.anchorterminal.com/compare/amazon-translate-vs-azure-translator.md ### 6. Google Cloud Translation, B 68.1/100 Google's API for translating text and documents. - Verdict: 500,000 free characters a month from a $10 credit, and only successful translations bill. Basic (v2) documents the API key as a URL query parameter. - Choose it for: Agents on Google Cloud that want high quotas, clear data terms and a choice of NMT, LLM or adaptive translation with glossaries. - Strength: 500,000 free characters a month from a $10 credit, and only successful translations bill - Strength: Content held in memory only, not used for training, not shared - Strength: 6 million characters a minute per project by default - Weakness: Basic (v2) documents the API key as a URL query parameter - Weakness: Quota errors come back as 403, not 429 - Weakness: No formality control, no MCP server and no llms.txt - Price: $80 / 1k pages · Auth: OAuth or key · x402: no · Where: hosted - Full assessment: https://www.anchorterminal.com/tools/google-cloud-translation.md - Against #1: https://www.anchorterminal.com/compare/azure-translator-vs-google-cloud-translation.md ### 7. Phrase, B 64.2/100 Localisation platform from Phrase a.s. in Prague. Its Language AI API translates text and files through automatically selected machine translation engines, and its Strings and TMS APIs, CLI and hosted MCP server manage translation projects, term bases and translation memories. - Verdict: Nine APIs with public OpenAPI specs, Markdown docs and a hosted MCP server that signs in with OAuth and separates view, create and delete consents. The status page lists 14 incidents in 90 days, 12 marked major or critical. Capacity top-up prices aren't public, and the terms permit training some models on customer content and prohibit benchmarking. - Choose it for: Teams that keep translation projects, term bases and translation memories in Phrase and want an agent to run them, or to translate through engines chosen per language pair. - Strength: Public OpenAPI specs for all nine APIs, an llms.txt index and a Markdown copy of every docs page, with a dated changelog of 51 entries since 24 April 2026 - Strength: Hosted MCP server in the EU and the US with OAuth, PKCE, dynamic client registration and three scopes, `phrase-mcp:read`, `phrase-mcp:write` and `phrase-mcp:destructive` - Strength: Service accounts since 7 August 2026 use the OAuth client credentials grant, limited by product and scope, and don't count as seats - Weakness: status.phrase.com lists 14 incidents between 10 July and 8 October 2026, two marked critical and ten major, most confined to one component - Weakness: The terms of 4 October 2026 let Phrase train non-text models on customer content without approval, and prohibit use for benchmarking - Weakness: The Strings API still documents email and password Basic authentication and an `access_token` query parameter - Price: $27 / mo · Auth: OAuth or key · x402: no · Where: hosted - Full assessment: https://www.anchorterminal.com/tools/phrase.md - Against #1: https://www.anchorterminal.com/compare/azure-translator-vs-phrase.md ### 8. Lingo.dev, B 62.3/100 Lingo.dev is a hosted AI translation platform from Replexica, Inc. Customers configure engines with glossaries, brand voices, rules and a model per language pair, then call them through a REST API, a CLI or a hosted MCP server. - Verdict: Lingo.dev suits teams that want glossary and tone control on every translation call, with a published fee of $2 per million output tokens and model costs passed through. It has no public OpenAPI file, its `llms.txt` is one line, and scoped keys, roles and audit logs are limited to the Enterprise plan. - Choose it for: Product teams that translate interface strings and content into many locales and want glossary, tone and rule control held in one place, driven from code, CI or a coding assistant. - Strength: Glossary terms, per-locale brand voice and rules apply on every call, sync or async, through one configured engine - Strength: Each sync response reports the model used, token counts and cost in USD - Strength: Async jobs take up to 100 target locales, an `idempotencyKey`, signed webhooks and cursor paging with `limit` up to 100 - Weakness: No public OpenAPI file was found, `llms.txt` holds one title line and docs pages have no Markdown version - Weakness: Organisation keys with their own role and engine scope, RBAC and audit logs are Enterprise only. On other plans every member is an Owner - Weakness: The status page lists two partial API outages in 90 days, on 18 July 2026 (closed after 19 hours) and 22 September 2026 (1 hour 50 minutes) - Price: $99 / mo · Auth: OAuth or key · x402: no · Where: hosted - Full assessment: https://www.anchorterminal.com/tools/lingo-dev.md - Against #1: https://www.anchorterminal.com/compare/azure-translator-vs-lingo-dev.md ### 9. Lara Translate API, C 61.3/100 Translated's translation model behind SDKs and an MCP server, with glossaries, translation memories, faithful, fluid or creative styles, free-text instructions, document translation and language detection. - Verdict: Glossaries, translation memories, styles and instructions on every translate call. Texts may be used for model improvement unless a request sets `noTrace`. - Choose it for: Agents that localise for one brand and need glossaries, translation memories and style instructions on every call, especially from an MCP client. - Strength: Glossaries, translation memories, styles and instructions on every translate call - Strength: Official MCP server, 22 tools with readOnlyHint, destructiveHint and idempotentHint, hosted with OAuth or local over stdio - Strength: Published per-million prices in dollars and euros, source characters only, detection free with translation - Weakness: Texts may be used for model improvement unless a request sets `noTrace` - Weakness: No public REST API or API changelog, so every call goes through an SDK or the MCP server - Weakness: No security.txt, disclosure policy or certification found - Price: Freemium · Auth: OAuth or key · x402: no · Where: hosted and local - Full assessment: https://www.anchorterminal.com/tools/lara-translate.md - Against #1: https://www.anchorterminal.com/compare/azure-translator-vs-lara-translate.md ### 10. LibreTranslate, D 49.8/100 Open-source machine translation API built on Argos Translate models. - Verdict: Open source under AGPL-3.0, self-hosted with no text leaving your network. No status page, SLA or terms of service for the hosted API. - Choose it for: Operators who must keep text on their own servers or want a fixed cost, and for light hosted use. - Strength: Open source under AGPL-3.0, self-hosted with no text leaving your network - Strength: Swagger 2.0 spec at /spec on every server - Strength: Hosted privacy policy says texts aren't stored or logged - Weakness: No status page, SLA or terms of service for the hosted API - Weakness: 2,000 characters a call and about 20 calls a minute sustained on Pro - Weakness: No glossaries or formality control, about 50 languages - Price: $29 / mo · Auth: API key · x402: no · Where: hosted - Full assessment: https://www.anchorterminal.com/tools/libretranslate.md - Against #1: https://www.anchorterminal.com/compare/azure-translator-vs-libretranslate.md 2 more are ranked in the full table: https://www.anchorterminal.com/categories/translation.md ## Head to head - [Azure Translator vs DeepL API](https://www.anchorterminal.com/compare/azure-translator-vs-deepl-api.md) - [Azure Translator vs Lokalise](https://www.anchorterminal.com/compare/azure-translator-vs-lokalise.md) - [Azure Translator vs Crowdin](https://www.anchorterminal.com/compare/azure-translator-vs-crowdin.md) - [Amazon Translate vs Azure Translator](https://www.anchorterminal.com/compare/amazon-translate-vs-azure-translator.md) - [DeepL API vs Lokalise](https://www.anchorterminal.com/compare/deepl-api-vs-lokalise.md) - [Crowdin vs DeepL API](https://www.anchorterminal.com/compare/crowdin-vs-deepl-api.md) - [Amazon Translate vs DeepL API](https://www.anchorterminal.com/compare/amazon-translate-vs-deepl-api.md) - [Crowdin vs Lokalise](https://www.anchorterminal.com/compare/crowdin-vs-lokalise.md) - [Amazon Translate vs Lokalise](https://www.anchorterminal.com/compare/amazon-translate-vs-lokalise.md) - [Amazon Translate vs Crowdin](https://www.anchorterminal.com/compare/amazon-translate-vs-crowdin.md) ## Questions ### What are the highest-rated machine translation APIs for AI agents? Azure Translator has the highest benchmark score of the 12 ranked machine translation APIs, 72 (BB). DeepL API is second with 72 (BB). ### How many machine translation APIs are agent-ready? 3 of the 12 ranked here grade BB or better, the bar for agent-ready on the Anchor benchmark. ### Which machine translation APIs accept x402 payments? None of the ranked listings here accepts x402 for its main call yet. ### How is this list ranked? By the Anchor benchmark score out of 100, a weighted mean of the scored categories minus deductions for negative events, from public evidence re-checked as vendors change. Listings cannot pay for a place. The latest assessment behind this page is from 8 October 2026. ## How this list is made The order is the Anchor benchmark score, the same number as on each listing. Each listing is graded from public evidence against the benchmark checklist, and the picks are worked out from those grades, prices and facts. No listing pays for its place, and paid audits or listing help never change a score.