VikhrModels/ru_llm_arena

Modified Arena-Hard-Auto LLM evaluation toolkit with an emphasis on Russian language

41
/ 100
Emerging

This tool helps evaluate how well different large language models (LLMs) perform when responding to prompts in Russian. It takes a list of LLM names and a set of 500 diverse Russian prompts, then automatically compares each LLM's answers against a baseline model (GPT-3.5-turbo-0125). The output is a clear ELO ranking and win-rate score for each LLM, showing which ones are best suited for generating Russian text. This is designed for researchers, AI product managers, and developers who need to select the most capable LLMs for Russian-language applications.

No commits in the last 6 months.

Use this if you need an automated, objective, and detailed comparison of various LLMs for their quality in generating responses in Russian.

Not ideal if you are evaluating English-language LLMs or if you prefer manual, human-based evaluation over automated metrics.

LLM-evaluation Russian-language-AI NLP-benchmarking AI-model-selection natural-language-generation
Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 8 / 25
Maturity 16 / 25
Community 17 / 25

How are scores calculated?

Stars

47

Forks

9

Language

Python

License

Apache-2.0

Last pushed

Mar 20, 2025

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/llm-tools/VikhrModels/ru_llm_arena"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.