README.md

---
pretty_name: HuggingFace AI Coding Tools Dashboard
task_categories:
  - text-generation
tags:
  - benchmark
  - ai-coding-tools
  - huggingface
language:
  - en
  - code
license: cc-by-4.0
size_categories:
  - 1K<n<10K
---

# HuggingFace AI Coding Tools Dashboard

Benchmark data from the [HuggingFace AI Dashboard](https://huggingface.submarine.ai) — tracking how AI coding tools (Claude Code, Codex, Copilot, Cursor) recommend HuggingFace products across 32 developer categories.

## Dataset Structure

| Split | Description | Rows |
|-------|-------------|------|
| `results` | Full benchmark results with LLM responses, cost, tokens, latency, and product detection | 8881 |
| `queries` | Benchmark query definitions across 32 categories | 263 |
| `runs` | Run metadata and tool/model configurations | 1 |
| `products` | HuggingFace product catalog with detection keywords | 44 |

## Key Fields (results)

- **tool**: AI coding tool tested (`claude_code`, `codex`, `copilot`, `cursor`)
- **model**: Specific model used
- **response**: Full raw LLM response text
- **detected_products**: HuggingFace products mentioned in the response
- **cost_usd / tokens_input / tokens_output / latency_ms**: Performance metrics

## Example Queries

**DuckDB:**
```sql
SELECT tool, COUNT(*) as mentions
FROM results
WHERE response LIKE '%xet%'
GROUP BY tool
```

**Python:**
```python
from datasets import load_dataset
ds = load_dataset("davidkling/hf-coding-tools-dashboard")
```