A recent independent test of 18 AI models for generating Russian text revealed that while top models like GPT-5.4 and Claude Opus 4.6 perform nearly identically, their pricing varies by a factor of 130. This significant cost difference, ranging from $0.0008 to $0.1014 per call, suggests that economic factors should heavily influence model selection for content creators. The test also highlighted specific issues with Russian text generation, such as the insertion of non-Cyrillic characters and the leakage of prompts into outputs, which can be mitigated with simple scripting. However, a more subtle form of AI-generated text, characterized by empty generalizations and unnecessary precision, still requires human fact-checking. AI
IMPACT Highlights the critical trade-offs between AI model quality and cost, emphasizing economic factors over marginal performance gains for content creation.
RANK_REASON Independent benchmark comparing multiple LLMs on a specific language task (Russian text generation) with detailed analysis of quality, cost, and defects. [lever_c_demoted from research: ic=1 ai=1.0]
- Claude
- Claude Opus 4.6
- DeepSeek V3
- DeepSeek V3.2
- GigaChat 3.5
- GigaChat 3 Ultra
- GigaChat Max
- GPT-4.1
- GPT-5
- GPT-5.2
- GPT-5.4
- MiMo V2 Omni
- Qwen3 235B
- YandexGPT 5.1 Pro
- YandexGPT Pro
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
