Jamba Reasoning 3B: benchmark results
AI21's Apache-2.0 3B reasoning model (October 2025) using the hybrid SSM-Transformer Jamba architecture, running 250K+ token contexts on-device. Provider: AI21 Labs. Released 2025-10-08.
Aggregate Rating 1433 ± 37, rank #4509 of 10916 rated models, from 25 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AA IFBench | 52.45 | Accuracy (%) | 62.7 |
| AA Omniscience - Software Engineering (SWE) - PHP | 16 | Accuracy (%) | 35 |
| AA Omniscience - Software Engineering (SWE) - Julia | 4 | Accuracy (%) | 30.4 |
| AA Omniscience - Software Engineering (SWE) - Swift | 24 | Accuracy (%) | 29.5 |
| AA LiveCodeBench | 20.95 | Pass@1 (%) | 22.1 |
| AA CritPt | 0 | Accuracy (%) | 20.9 |
| AA MMLU-Pro | 57.75 | Accuracy (%) | 20.7 |
| AA Omniscience - Software Engineering (SWE) - Go | 8 | Accuracy (%) | 20 |
| AA Omniscience - Software Engineering (SWE) - Python | 9 | Accuracy (%) | 16.2 |
| Tau2-Bench Telecom | 15.79 | Success Rate (%) | 14.2 |
| AA AIME 2025 | 10.67 | Accuracy (%) | 14 |
| AA Omniscience - Software Engineering (SWE) - R | 2 | Accuracy (%) | 14 |
Interactive version: theaggregate.ai/model?slug=jamba-reasoning-3b · How It Works · Data refreshed daily, snapshot 2026-11-08.