{"id":96,"date":"2026-09-23T11:24:41","date_gmt":"2026-09-23T11:24:41","guid":{"rendered":"https:\/\/docs.latw.ai\/uncategorized\/ai-models\/"},"modified":"2026-09-23T11:24:41","modified_gmt":"2026-09-23T11:24:41","slug":"ai-models","status":"publish","type":"post","link":"https:\/\/docs.latw.ai\/pl\/latw-for-polylang\/ai-models\/","title":{"rendered":"AI Models and Cost"},"content":{"rendered":"

You bring your own API key, and the engine bills you directly at its published rates \u2014 the plugin
\nadds nothing on top and never handles your money.<\/p>\n

Which engine and model you pick under AI Translation \u2192 Settings \u2192 General<\/strong> decides both the
\nquality and the bill. The engines themselves are described in Translation Engine – AI Providers<\/em>.<\/p>\n

Models per engine<\/h2>\n\n\n\n\n\n\n\n\n\n\n
Engine<\/th>\nFree version<\/th>\nAdded by PRO<\/th>\n<\/tr>\n<\/thead>\n
OpenAI<\/td>\nGPT-5.4 Nano (default)<\/em>, GPT-5.6 Luna<\/td>\nGPT-5.6 Terra, GPT-5.6 Sol, GPT-5 Nano, GPT-5 Mini, GPT-5, GPT-5.4, GPT-5.4 Mini, GPT-5.5<\/td>\n<\/tr>\n
Google Gemini<\/td>\nGemini 3.5 Flash Lite<\/td>\nGemini 3.7 Flash, Gemini 3.6 Flash, Gemini 3.5 Flash<\/td>\n<\/tr>\n
Google Translate<\/td>\nNeural Machine Translation, Phrase-Based (legacy)<\/td>\n\u2014<\/td>\n<\/tr>\n
MyMemory<\/td>\nno model to choose<\/td>\n\u2014<\/td>\n<\/tr>\n
Anthropic Claude<\/td>\n\u2014<\/td>\nClaude Opus 5, Claude Sonnet 5, Claude Haiku 4.5<\/td>\n<\/tr>\n
DeepL<\/td>\n\u2014<\/td>\nPrefer quality (recommended)<\/em>, Quality optimised, Latency optimised<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

The free version ships the cheapest models of each engine. PRO adds the rest, for content where the
\ncheap models struggle \u2014 marketing copy that has to land, legal or medical text, anything where tone
\nmatters more than throughput.<\/p>\n

Reasoning effort<\/h2>\n

Settings \u2192 General \u2192 Reasoning Effort<\/strong> controls how much thinking a language model does before
\nanswering. Higher effort improves quality on difficult passages, and costs more and takes longer.
\nLow is the default and is the right choice for most content. Machine-translation engines ignore it.<\/p>\n

What it actually costs<\/h2>\n

Language models are billed per token \u2014 roughly, per chunk of text in and out. As a rough guide, a
\nsingle 1,000-word page through GPT-5.4 Nano costs well under one cent. Google Translate and DeepL
\nare billed per character sent. MyMemory is free within its daily quota.<\/p>\n

You do not have to rely on estimates. Every finished job records what it actually cost, priced
\nagainst the model that produced it, and AI Translation \u2192 History<\/strong> shows real figures for your own
\ncontent. The Dashboard totals the last 30 days.<\/p>\n

Cost history and PRO<\/h2>\n

Prices for the PRO models are kept even when the licence is not active, deliberately.<\/p>\n

If your licence lapses, the PRO models disappear from the model list \u2014 but the cost already recorded
\nfor translations made with them stays correct. Cost history is a record of what you spent, not a
\nfeature to be switched off.<\/p>\n

Keeping the bill down<\/h2>\n
    \n
  • Write a website description.<\/strong> Better context means fewer retranslations, which is the cheapest
    \nsaving available.<\/li>\n
  • Use Batch processing<\/strong> for work that does not have to be done now: about half the price, on
    \nOpenAI and Claude. See Translation Engine – AI Providers<\/em>.<\/li>\n
  • Update, do not re-translate.<\/strong> With PRO, editing an already translated post and translating it
    \nagain sends only what changed. See Updating Outdated Translations<\/em>.<\/li>\n
  • Translate what you need.<\/strong> Target languages are chosen per translation, not set once globally.<\/li>\n
  • Start on the cheapest model<\/strong> and move up only for the content that visibly needs it. There is
    \nno reason to pay reasoning-model prices for a page of product specifications.<\/li>\n
  • Set a spending limit on your provider account.<\/strong> The plugin cannot do this for you, and it is
    \nthe only hard stop that exists. For automatic translation, PRO adds daily limits of its own \u2014 see
    \nAutomatic Translation<\/em>.<\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"

    You bring your own API key, and the engine bills you directly at its published rates \u2014 the plugin adds nothing on top and never handles\u2026<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_latwm_translated_slug":"","footnotes":""},"categories":[17],"tags":[],"class_list":["post-96","post","type-post","status-publish","format-standard","hentry","category-latw-for-polylang"],"_links":{"self":[{"href":"https:\/\/docs.latw.ai\/pl\/wp-json\/wp\/v2\/posts\/96","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/docs.latw.ai\/pl\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/docs.latw.ai\/pl\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/docs.latw.ai\/pl\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/docs.latw.ai\/pl\/wp-json\/wp\/v2\/comments?post=96"}],"version-history":[{"count":0,"href":"https:\/\/docs.latw.ai\/pl\/wp-json\/wp\/v2\/posts\/96\/revisions"}],"wp:attachment":[{"href":"https:\/\/docs.latw.ai\/pl\/wp-json\/wp\/v2\/media?parent=96"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/docs.latw.ai\/pl\/wp-json\/wp\/v2\/categories?post=96"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/docs.latw.ai\/pl\/wp-json\/wp\/v2\/tags?post=96"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}