Ru Bfcl: Berkeley Function Calling Leaderboard Evaluation Dataset

A live leaderboard dataset for evaluating the function calling capabilities of large language models. The dataset was created by AvitoTech to represent common user scenarios, such as in AI agents and enterprise workflows. It spans diverse categories and multiple languages, with the last recorded update on June 4, 2026.

Use Cases

• Benchmarking LLM performance on function calling tasks based on the described evaluation categories.
• Training or fine-tuning models for agentic workflows based on the representative use-cases mentioned.
• Comparing multilingual function calling capabilities across different models based on the dataset's language coverage.

Strengths

• Designed to be representative of common function calling use-cases like agents and enterprise workflows.
• Covers diverse categories and multiple languages, as stated in the description.
• Maintained as a live leaderboard, with a last recorded update on 2026-06-04.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and file format are unknown, which may limit suitability assessment.
• Data may reflect bias inherent to the specific evaluation tasks and sources used.

Provenance

Source

AvitoTech, hosted on Hugging Face.

Collection Method

Built from learnings to represent common function calling use-cases.

Freshness

Last updated 2026-06-04 17:01:40; freshness should be verified.