Are the AI translations included in my subscription?
Yes. The built-in AI and machine translation services are part of every paid plan; there is no separate contract or extra subscription. Usage is powered by credits:
- AI tokens power the Locize AI translation service (and Quality Estimation). Tokens cover both input (source text, glossary, styleguide) and output (translated text and any reasoning tokens the model spends).
- MT characters power the classic machine translation service, billed per source character processed.
What is included
- Every paid plan includes an AI-token and MT-character allowance with each billing cycle; the included amounts per plan are listed on the pricing page.
- Every new project includes starter credits, so automatic translation already works during the trial.
- Usage-based subscriptions receive bonus credits with each paid invoice instead of a fixed monthly amount.
You can see your remaining credits directly in the app: on the project overview (Project health card) and on your plan page. The number shown there is the current balance of your subscription (included credits plus top-ups, minus usage), not a fixed plan limit, so it changes over time.
What if the included credits are used up?
You can top up additional AI tokens or MT characters directly in the app; they are simply added to your next regular invoice. The per-token and per-character prices are listed on the pricing page.
MT characters vs. AI tokens
The two units are different, so comparing a "price per 100k" across them is misleading:
- MT characters are billed per source character. The built-in Locize MT runs on Google Translate, which is priced per character, and that per-character rate is far above what large language models charge per token. The Locize MT price reflects this upstream cost.
- AI tokens are billed per token (roughly 4 characters of English text, fewer for other scripts). Tokens count both the input, meaning the source text plus the context Locize sends along (instructions, glossary, styleguide, key context), and the output, meaning the translated text plus any reasoning tokens the model spends.
Per translated segment the two usually end up in the same range. Example for a 60-character UI string at list prices:
| Consumes | Costs about (USD) | |
|---|---|---|
| Locize MT | 60 characters | 0.002 |
| Locize AI | 200 to 800 tokens, most of it context | 0.0016 to 0.0064 |
Very short strings are cheapest through MT. Longer texts tilt the balance towards AI, because the context overhead is spread over more text, and the automatic translation workflow translates new keys in batches, which spreads it over several keys. In most projects quality decides, not price: AI takes glossary, styleguide and key context into account and its output is scored by Quality Estimation, while MT translates the text as it is. See Using AI translation effectively.
Using your own AI provider instead
On the Growth plan and above you can bring your own key (OpenAI, Gemini, Mistral, DeepL, ...). Usage then runs through your own provider account and does not consume Locize AI tokens. Learn more →
Running several projects? Store the key once for your organization (organization page → Settings → MT/AI credentials) and let each project use it with a single switch in its provider configuration.