Falcon3-1B-Base: benchmark results
TII's Falcon3-1B-Base, pruned and healed from a larger 3B Falcon model via knowledge distillation on 80 billion tokens, with a 4K context and support for English, French, Spanish and Portuguese. Provider: TII. Released 2024-12-17. Access: Open.
Aggregate Rating 1215 ± 28, rank #1549 of 1635 rated models, from 83 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard - MuSR | 41.47 | Score | 54.9 |
| Open LLM Leaderboard - GPQA | 27.94 | Score | 34.2 |
| Open Portuguese LLM - FaQuAD NLI | 50.06 | Macro F1 (%) | 32 |
| Swallow - Pre-trained English - GSM8K | 33.7 | Exact Match (%) | 30.6 |
| Swallow - Pre-trained English - OpenBookQA | 31.6 | Accuracy (%) | 26.5 |
| Open Japanese LLM - Jsick Exact Match | 62.09 | Score (%) | 24.3 |
| Swallow - Pre-trained English - MATH | 14 | Exact Match (%) | 23.5 |
| Open Portuguese LLM - OAB Exams | 28.06 | Accuracy (%) | 23.3 |
| Open Portuguese LLM - ENEM | 27.99 | Accuracy (%) | 22.7 |
| Open Portuguese LLM - BLUEX | 26.01 | Accuracy (%) | 22.2 |
| Swallow - Pre-trained English - MMLU | 44.9 | Accuracy (%) | 20.4 |
| Open LLM Leaderboard - BBH | 35.71 | Score | 20.1 |
Interactive version: theaggregate.ai/model?slug=falcon3-1b-base · How It Works · Data refreshed daily, snapshot 2026-11-05.