By the end of this page you will know exactly what DeepL, Phrase and Lokalise charge in 2026, which two of them removed or gated their entry tiers in the last twelve months, what a 30-person Hong Kong team actually pays per year, and the one quality problem that matters more than price if your content is in Traditional or Simplified Chinese.
This is a buying decision most Hong Kong enterprises make badly, because they price the tool and never price the workflow. Every figure below is a published vendor rate or a dated archive of one, current as of August 2026.
How much does DeepL cost in 2026?
DeepL Translator runs from US$0 to US$57.49 per user per month on annual billing. The tiers are Free, Individual at US$8.74, Team at US$28.74, Business at US$57.49, and Enterprise on a custom quote. Monthly billing costs roughly 16% more across every plan.
DeepL Translator plans and limits, August 2026
--- Free: US$0, 50,000 characters per month, 1 file translation up to 5 MB, 1 glossary with 5 entries, no credit card
--- Individual: US$8.74 per month annual, 300,000 characters, 3 file translations up to 30 MB, 1 glossary, 30-day trial
--- Team: US$28.74 per user per month annual, 1,000,000 characters per user, 20 file translations, 5 shared glossaries at 1,000 entries per language pair, analytics, SSO for teams of 50 or more
--- Business: US$57.49 per user per month annual, unlimited characters under fair usage, 100 file translations, translation memory, SCIM provisioning, 99.0% availability SLA, Write Pro included
--- Enterprise: custom pricing, 99.9% SLA, bring-your-own encryption key, dedicated account team
--- Write Pro add-on: US$7.49 per user per month on Individual and Team, included on Business and Enterprise
The DeepL API is a separate product with separate billing. The free Developer tier gives 1 million characters in total, once, not per month. Growth costs US$26 per month on annual billing and includes 12 million characters per year, with overage at US$27.50 per additional million characters.
The word "unlimited" on Business and Enterprise carries a fair usage policy rather than a hard published cap. There is no stated ceiling, but sustained multi-million-character volume can trigger contact from DeepL.
What happened to Phrase pricing?
Phrase removed its entry tier. The US$135 per month Starter plan, introduced in February 2024, disappeared from the public pricing page between August and October 2025. By mid-2026 the entry business plan, Team, sits at US$1,245 per month billed annually, roughly US$15,000 per year, behind a "Get in touch" button.
What remains self-serve as of July 2026 is a US$525 per month Software UI/UX developer plan and a US$27 per month Freelancer plan aimed at individual translators with a single seat.
For a product or marketing team, the effective self-serve entry moved from US$135 to US$525, and the negotiated business floor now sits near US$15,000 a year.
This is not product decay. Forrester named Phrase a Leader in its first Translation Management Systems Wave in Q3 2025. It is a deliberate choice of customer. If you run a multi-vendor localisation programme with language service provider workflows, Phrase's Business and Enterprise tiers may genuinely be the right shelf. The issue is that everyone below that shelf now needs a different answer.
What changed at Lokalise in the November 2025 restructure?
Lokalise replaced its long-standing Start, Essential and Pro plans, priced at US$120, US$230 and US$825 per month, with a four-tier structure, moved billing from seats to processed words, withdrew its free-forever plan, and put two of the four tiers behind a demo request.
Lokalise pricing after the restructure
--- Explorer: US$144 per month
--- Growth: listed at US$499 per month at launch, US$375 per month as of July 2026
--- Advanced: US$999 per month, behind "Book a demo"
--- Enterprise: custom, behind "Book a demo"
Three structural changes came with the new names. Billing switched to processed words, so heavy machine translation usage now meters against you. The free plan, introduced only months earlier in 2025, was withdrawn, leaving a 14-day trial. And AI translation, previously a paid add-on at roughly US$0.05 per word, is now bundled into every plan.
Net effect: entry pricing rose about 20%, the self-serve ceiling fell from US$825 to the US$375 to US$499 range, and half the product line now requires a sales conversation.
What does a Hong Kong team actually pay per year?
A five-person content team on DeepL Team pays US$143.70 per month, or US$1,724 per year on annual billing. A 30-person enterprise on DeepL Business pays US$1,724.70 per month, or US$20,696 per year. A team that needs a full translation management system starts at US$4,500 a year on Lokalise Explorer or US$15,000 a year on Phrase Team.
Worked annual totals, annual billing, August 2026 list rates
--- DeepL Individual, 1 seat: US$104.88 per year, or US$194.76 with Write Pro
--- DeepL Team, 5 seats: US$1,724 per year, or roughly US$2,173 with Write Pro for all five
--- DeepL Business, 30 seats: US$20,696 per year, Write Pro included
--- DeepL Enterprise, 30 seats: custom, typically budgeted north of US$25,000 per year at that headcount
--- Lokalise Explorer: US$1,728 per year; Growth at US$375 per month is US$4,500 per year
--- Phrase Software developer plan: US$6,300 per year; Phrase Team is roughly US$15,000 per year
--- Crowdin Pro: from US$50 per month on annual billing, one of the few majors that kept transparent self-serve pricing
--- Smartling: no published platform subscription price; enterprise tiers are quoted against volume and workflow
Note what these numbers do not include: the internal hours to review output, maintain glossaries, and keep terminology consistent across three languages. In most Hong Kong organisations that internal cost exceeds the licence cost within the first year.
Why did both vendors move upmarket at the same time?
The localisation industry has stopped growing the way it used to, and the sub-US$500 per month self-serve segment is the part large language models absorb first. Funded vendors are concentrating on enterprise contracts where governance and workflow complexity still command budgets.
Nimdzi's 2026 industry report projects under 1% annual growth for the coming years and records clients expecting AI-driven cost reductions of up to 75%. CSA Research measured the industry's first outright revenue decline in 2023. Consolidation has followed, with XTM International acquiring Transifex in January 2025.
At the same time, the smallest projects increasingly skip a translation management system entirely and run their strings through a model in a continuous integration pipeline. That pinches the low end hardest.
The strategic read for a Hong Kong department head is that the sub-US$500 tier is not being competed for. It is being vacated. If your organisation sat comfortably in that band, your renewal is not a negotiation problem. It is a category problem, and the answer is more likely a workflow change than a cheaper vendor.
Is DeepL the right choice for Traditional and Simplified Chinese?
This is the question that should decide the purchase for most Hong Kong organisations, and price does not answer it. DeepL's quality advantage is strongest in European language pairs. For East Asian languages the picture is materially more mixed, and the evidence does not favour it.
A structured comparison run by Tyson Batino, testing five tools with native Japanese speakers, ranked DeepL last, behind ChatGPT, Gemini, Claude and Google Translate. Multiple user reports describe similar gaps in Chinese and Korean.
The cost gap compounds the problem. A developer comparison found real-time translation through the DeepL API costing around US$4.05 per hour against under US$0.01 per hour for a fast general-purpose model, a difference of roughly two orders of magnitude. For teams where translation is a commodity workflow rather than the core product, that maths is moving quickly.
Where DeepL keeps a defensible edge is document translation with formatting preserved, glossary-enforced terminology, and enterprise compliance including SOC 2 Type 2, ISO 27001 and bring-your-own encryption key on Enterprise. If those matter to your workflow, the premium is justifiable. If you mainly need bulk Chinese text translated well, test before you commit.
What are the honest limitations of every option here, including ours?
No option on this page wins every row, and any comparison that says otherwise is an advertisement. Each of these tools has a specific buyer it serves badly, and that includes what UD does.
DeepL: "unlimited" is fair-usage bounded with no published cap; East Asian language quality is contested; customer support scores 4.2 out of 5 on Capterra with a consistent complaint pattern around slow API issue resolution.
Phrase: below roughly US$15,000 a year you are buying a developer plan, not the product the Forrester Wave rated. If you do not run multi-vendor language service provider workflows, you will pay for governance you never use.
Lokalise: processed-word billing punishes exactly the machine-translation-heavy workflow the bundled AI encourages. Model your word volume before signing, not after.
Smartling: no published price makes budget planning and board approval harder, and the quote depends on volume you may not be able to forecast in year one.
UD: we are not a reseller of any platform named here and we hold no vendor discounts, so we cannot beat a direct purchase on licence price. If your requirement is a single-tool licence with no workflow redesign and no terminology governance, buy directly from the vendor and skip an implementation partner. Where we earn our fee is when the content operation itself, three languages, brand terminology, review cycles and publication, is the thing that needs building.
What is the right next step?
Do not start by comparing licence prices. Start by measuring the volume, the language pairs and the review burden you actually carry each month, because those three numbers decide which tier you need and whether a translation management system is warranted at all.
If you publish under 100,000 characters a month in two languages, a DeepL Individual or Team seat plus a disciplined glossary is almost certainly enough, and a translation management system is overhead you do not need yet.
If you publish across three languages with brand terminology, approval workflows and a website that has to stay in sync, you are buying a content operation rather than a translation tool, and the licence is the smaller half of the cost.
Every organisation that gets this wrong makes the same mistake: it buys the tool first and designs the workflow afterwards. Twenty-eight years of Hong Kong enterprise work says the sequence matters more than the vendor. We understand AI. We understand you. With UD by your side, AI never feels cold.
Reviewed by the UD enterprise AI team, Hong Kong. Vendor pricing reflects published rates and dated archive snapshots as of August 2026; verify current pricing with each vendor before purchase.
Related reading: How Much Does Enterprise AI Actually Cost? A Framework for Managing AI Spend and AI Writing Tools Compared: What Jasper, Copy.ai and Canva Really Cost.
Where to start
If your multilingual content is a recurring operation rather than a one-off project, the licence is the easy part. We'll walk you through every step, from mapping your actual volume and language pairs to designing the review workflow, standing up the terminology governance, and running the content operation itself with an AI Staff marketing employee.
