{"slug": "translating-1m-tokens-into-russian-on-gpt-5-6-costs-16-88-not-the-2-you", "title": "Translating 1M tokens into Russian on GPT-5.6 costs $16.88, not the $2 you estimated", "summary": "A developer measured that translating one million English tokens into Russian on GPT-5.6 costs $16.88, not the $2.00 a naive input-only estimate suggests, an 8.44x gap driven by output tokens making up 88% of the bill. Using a 608-word English/Russian parallel catalog and the o200k_base tokenizer, the analysis found Russian needs 1.24x more tokens than English, and that the same documents cost 44% more to translate English-to-Russian than in reverse. The writeup provides a cost formula and advises comparing models on output price, since a 10% cheaper output price cuts the bill by 8.8% versus 1.2% for input.", "body_md": "Say you want to translate your app's documentation into Russian with an LLM. You count the tokens in the docs, say a million, look up the price, $2 per million on GPT-5.6, and budget $2. Then the bill comes in at about $17.\n\nNothing went wrong. The estimate only counted the cheaper half of the work.\n\nA **token** is a chunk of text the model reads as one piece; APIs bill per million of them. **Input** tokens are what you send, and **output** tokens are what the model writes back. Output usually costs five or six times more per token than input. A **tokenizer** is the program that splits text into tokens; I used `tiktoken` with `o200k_base`, the encoding of the GPT-5.x models. A **parallel text** is the same content written in two languages, sentence by sentence.\n\nIn most LLM tasks the answer is much shorter than the prompt. A summary of a 10-page report is a paragraph, and a classifier answers with one word. Translation is different: the model has to write the entire document again, in another language. So the output is roughly as long as the input, and every one of those tokens is billed at the higher price.\n\nTo get real numbers I needed a real parallel text. The catalog of the LLM gateway I work on has a short description of each of its 50 chat models, written separately in English and in Russian. That's 608 English words and 620 Russian words of normal product text. In `o200k_base`, the English side is 1,105 tokens and the Russian side 1,366, so the same meaning takes 1.24x more tokens in Russian.\n\nTranslating English to Russian means sending 1,129 tokens (the text plus a 24-token instruction) and receiving 1,366. Scaled to a million English source tokens, with official list prices:\n\n| Model | Input / output $ per 1M | Naive estimate | Real bill | Multiplier | Output share | \n|---|---|---|---|---|---|\n| gpt-5.6 | 2 / 12 | $2.00 | $16.88 | 8.44x | 88% | \n| gpt-5.6-mini | 0.20 / 1.20 | $0.20 | $1.69 | 8.44x | 88% | \n| claude-sonnet-5.5 | 2 / 10 | $2.00 | $14.41 | 7.20x | 86% | \n| gemini-3.8-flash | 0.75 / 3.75 | $0.75 | $5.40 | 7.20x | 86% | \n| grok-4.7 | 2 / 6 | $2.00 | $9.46 | 4.73x | 78% | \n\nThe naive estimate is wrong by 5x to 8x, and how wrong depends on one number: how much more the model charges for writing than for reading. GPT-5.6 and its mini version charge 6x for output, the Claude and Gemini models here 5x, Grok 4.7 only 3x. Grok and GPT-5.6 have the same input price, but the real bill differs by 1.8x.\n\nThe multiplier doesn't need a spreadsheet. It's one plus the ratio of output tokens to input tokens, times the ratio of output price to input price:\n\n``` python\ndef translation_cost(source_tokens, in_price, out_price, length_ratio):\n    \"\"\"Prices in $ per 1M tokens; length_ratio = target tokens / source tokens.\"\"\"\n    output_tokens = source_tokens * length_ratio\n    return (source_tokens * in_price + output_tokens * out_price) / 1e6\n\n# English -> Russian on gpt-5.6: 1 + 1.24 * 6 = 8.44x the naive estimate\nprint(translation_cost(1_000_000, 2.00, 12.00, 1.24))  # 16.88\n```\n\nThe `length_ratio` depends on the language pair and the tokenizer. For this text it was 1.24 into Russian. In the other direction it's 0.81, because English is shorter: Russian to English on GPT-5.6 comes out at $11.74 per million source tokens, 5.87x the naive estimate and 83% output. The same documents cost 44% more to translate one way than the other. To get your own ratio, translate a few pages and count both sides with the tokenizer.\n\nBecause output is 80–90% of a translation bill, the output price is what you should compare between models. Check the input price last. On GPT-5.6, English to Russian, a 10% cheaper input price cuts the bill by 1.2%. A 10% cheaper output price cuts it by 8.8%, and so does 10% less output.\n\nLess output is something you control. Tell the model to return only the translation, without \"Here is the translated text:\" or notes at the end. Don't send text that doesn't need translating, like code blocks, URLs or tables of numbers; skip them and put them back in your own code. If the model does reasoning, turn it off or set it to the minimum for plain translation, because reasoning tokens are billed as output too, even though you never see them.\n\nThis is one small text: 50 product descriptions, about 600 words per language. Documentation with code, long sentences or a lot of terminology will have a different length ratio, so measure your own before you budget.\n\nI counted all tokens with OpenAI's tokenizer. Claude, Gemini and Grok use their own tokenizers, so their rows show the price effect only; their real token counts, and the length ratio, will be somewhat different.\n\nI didn't measure translation quality. The cheapest row in the table isn't automatically good enough, and a model that writes a looser, shorter translation is cheaper partly because it leaves things out.\n\nAnd the formula is a floor. It assumes one clean pass. Retrying a failed chunk, sending a glossary with every call, or asking for a second \"review\" pass all add to the bill, and the review pass alone writes the whole document a third time.", "url": "https://wpnews.pro/news/translating-1m-tokens-into-russian-on-gpt-5-6-costs-16-88-not-the-2-you", "canonical_source": "https://dev.to/altrouter/translating-1m-tokens-into-russian-on-gpt-56-costs-1688-not-the-2-you-estimated-4o1m", "published_at": "2026-11-18 12:00:02+00:00", "updated_at": "2026-11-18 12:13:52.158573+00:00", "lang": "en", "topics": ["large-language-models", "natural-language-processing", "ai-tools"], "entities": ["GPT-5.6", "GPT-5.6-mini", "Claude Sonnet 5.5", "Gemini 3.8 Flash", "Grok 4.7", "tiktoken"], "also_reported_by": [], "takeaway": "Budget LLM translation at roughly 1 + length_ratio × (output price / input price) times the naive input-only estimate — 8.44x for English-to-Russian on GPT-5.6, where output is 88% of the bill.", "recommendation": {"advice": "Estimate translation cost with the output-length ratio and output price, not input price alone, and instruct the model to return only the translation to cut output tokens.", "playbook": "llm-cost", "playbook_url": "https://wpnews.pro/playbooks/llm-cost"}, "alternates": {"html": "https://wpnews.pro/news/translating-1m-tokens-into-russian-on-gpt-5-6-costs-16-88-not-the-2-you", "markdown": "https://wpnews.pro/news/translating-1m-tokens-into-russian-on-gpt-5-6-costs-16-88-not-the-2-you.md", "text": "https://wpnews.pro/news/translating-1m-tokens-into-russian-on-gpt-5-6-costs-16-88-not-the-2-you.txt", "jsonld": "https://wpnews.pro/news/translating-1m-tokens-into-russian-on-gpt-5-6-costs-16-88-not-the-2-you.jsonld"}}