AI lab for localization

Not on the homepage. Open Pricing in the footer, or Try platform → Plans. Both go to /pricing.

That page has text plans (Free, Standard, Pro, Premium, Enterprise), a separate voice package sold by the minute, and an estimator: pick a monthly word count and it names the plan.

Every plan uses the same Algebras Agent. The difference is volume, languages, and support. Quality does not change with the tier. Annual billing is two months free. Overage is per 1,000 words of translated text.

The pricing page lists Standard as 30 languages and Pro as 300 languages, plus advanced glossary tools and multi-language QA. Premium adds API access, glossaries, a personal manager, and custom integrations. Enterprise adds custom models, benchmarking, a dedicated manager, and an SLA.

Voice is priced apart from words: 10 minutes for $15, 60 minutes for $79, 240 minutes for $240. Listener hours and concurrency limits sit on the same Pricing page.

Yes. Try for free and Start a free project both open platform.algebras.ai. The homepage says no card is required. The free tier on Pricing covers up to 1,000 words, for a test project.

Log in later at platform.algebras.ai/auth/sign-in.

Words in the finished translation, not tokens, characters, or API calls. Only a completed translation counts. The dashboard shows words left, usage, and a projected monthly cost, and it warns you before you hit the limit.

If you pass the limit, translation does not stop and the account is not locked. You get an alert and an instant upgrade that keeps the project. You can also move from Free to Standard yourself. Cancel anytime; the plan runs to the end of the period already paid, with no penalty.

The open Translation Benchmark, linked from Research in the header: benchmark.algebras.ai. It ranks models by fluency and by cost across 300 language pairs, 30+ models, and 8 source languages. FluencyScore itself, the five-dimension naturalness check, is at benchmark.algebras.ai/fluency.

On a paid plan, one Fluency API request or batch item counts as 10 words.

A self-serve project starts at platform.algebras.ai. From a coding agent, Start localizing and the October 2026 announcement both go to the Agentic Localization Harness at /agent-harness, and View on GitHub goes to github.com/algebras-ai/algebras-agent. You install it in Claude Code, Cursor, Codex, Windsurf, or Copilot and say “Translate this project.”

A walkthrough with the team is Book a demo in the header (a Calendly link). Talk to our team and Contact sales both go to /contact. Email is hello@algebras.ai.

The homepage stack is four doors. For vibecoding is the harness and MCP server, /agent-harness. For games is /industries/ai-game-localization-agent. Voice & video (dubbing, lip sync, subtitles) is /industries/voice-video-localization. For developers is the CLI at /web-localization-cli; the API docs are docs.algebras.ai, and the OpenAPI spec is platform.algebras.ai/api/v1.

The icon row lower on the page repeats the same tools: CLI, API, Unity, Gridly, and GitHub Actions each open a guide. Unreal and TMS are on the page but not linked to a guide yet.

Docs and the API live in the footer (docs.algebras.ai and the API). Help desk is help.algebras.ai. Service status is status.algebras.ai, under Research. The blog, Notes from the lab, is /blog; the email field on the homepage is the monthly note, not a support form.

Case studies are on the homepage and in full at /cases: Epoint, Evolve, PlayDucky, Mixed Realms, InDrive, and Mama Space. Research papers and the dataset notes are under Research → open.algebras.ai.

The site language switch is EN, RU, and ZH in the header. Prices and this FAQ are in English.