Within the same language, MT engine solutions cost ~5 times as much as LLM solutions at the median and have more errors.
50% of the one-step solutions cost from $3.17 to $21.05 per million characters, and the median one costs $6.98. At $0.16 to $0.24 per word, the human translation we priced in the 2025 edition comes to ~$19,000 to $28,000 per million characters.
75% of the LLM solutions cost under $10 per million characters, and 99% of the MT engine solutions cost $10 or more. The cheapest MT engine solutions, at $3.94 per million characters on Yandex, cost more than the LLM solutions at the median.
Cost per million characters and weighted errors per 1,000 words of the one-step solutions, by model
| Model | Type | Cost per million characters, median | Weighted errors per 1,000 words, median | Meeting every Standard requirement | Languages |
|---|---|---|---|---|---|
| Anthropic Claude Fable 5 | LLM | $32.28 | 19.4 | 35% | 60 |
| DeepL Next Gen | LLM-based MT engine | $27.50 | 42.8 | 0% | 71 |
| Translated Lara 2 | LLM-based MT engine | $24.99 | 38.2 | 0% | 44 |
| Google Translation LLM | LLM-based MT engine | $21.32 | 81.9 | 0% | 32 |
| Google Cloud Advanced Translation API | MT engine | $20.00 | 121.3 | 0% | 4 |
| Anthropic Claude Opus 4.8 | LLM | $15.83 | 25.4 | 7% | 46 |
| Amazon | MT engine | $15.00 | 76.1 | 0% | 7 |
| OpenAI GPT-5.5 | LLM | $11.61 | 22.8 | 24% | 62 |
| Microsoft Translator | MT engine | $10.00 | 67.9 | 0% | 32 |
| Widn AI Tower Sugarloaf | LLM-based MT engine | $10.00 | 60.2 | 0% | 10 |
| OpenAI GPT-5.6 Sol | LLM | $7.90 | 18.8 | 24% | 33 |
| Anthropic Claude Sonnet 4.6 | LLM | $7.37 | 35.6 | 0% | 31 |
| Anthropic Claude Sonnet 5 | LLM | $5.98 | 35.3 | 0% | 35 |
| Google Gemini 3.1 Pro Preview | LLM | $4.76 | 25.2 | 21% | 76 |
| OpenAI GPT-5.6 Terra | LLM | $4.56 | 20.6 | 25% | 20 |
| Google Gemini 3.5 Flash | LLM | $3.57 | 35.9 | 5% | 64 |
| OpenAI GPT-4.1 | LLM | $3.06 | 40.2 | 2% | 45 |
| z.ai GLM-5 | LLM | $2.40 | 43.2 | 3% | 33 |
| xAI Grok 4.20 (non-reasoning) | LLM | $1.28 | 67.6 | 0% | 9 |
| MoonshotAI Kimi K2.5 | LLM | $1.26 | 44.7 | 0% | 12 |
| xAI Grok 4.20 (reasoning) | LLM | $1.23 | 43.5 | 0% | 16 |
| DeepSeek V3.2 | LLM | $1.21 | 52.2 | 0% | 14 |
| Mistral Medium 3.1 | LLM | $1.02 | 57.0 | 0% | 4 |
| Mistral Large 3 | LLM | $0.82 | 45.3 | 0% | 28 |
| OpenAI GPT-5.6 Luna | LLM | $0.45 | 28.9 | 12% | 26 |
| Google Gemma 4 | Open-weight LLM | $0.14 | 47.8 | 3% | 37 |
Units — Cost per million characters, median: US$ per million characters; Weighted errors per 1,000 words, median: weighted errors per 1,000 words; Meeting every Standard requirement: % of the languages
One-step solutions on the shortlist with a cost and a conclusive Intento LQA result, models with three or more such languages. Errors weighted by severity: minor 1, major 5, critical 10.