Jamba 1.5 Mini: benchmark results
AI21's hybrid Mamba-Transformer MoE (52B total, 12B active) with a 256K context. Provider: AI21 Labs. Released 2024-08-22. Access: API.
Unified ELO 1456 ± 1, rank #918 of 1392 rated models, from 33 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| RULER | 93.9 | Avg Accuracy (%) | 87.5 |
| HELMET (128K) | 46.9 | Average Score | 87.1 |
| NoLiMa | 92.4 | Base Score (%) | 66.7 |
| HELM NarrativeQA | 74.62 | F1 (%) | 57.2 |
| HELM NaturalQuestions (Closed) | 38.79 | F1 (%) | 54.4 |
| HELM NaturalQuestions (Open) | 71.02 | F1 (%) | 47.8 |
| HELM Lite | 45.83 | Mean win rate (self-reported) | 44.7 |
| HELM (Stanford) | 41.38 | Mean Win Rate (%) | 38.9 |
| HELM WMT 2014 | 17.9 | BLEU-4 (%) | 37.8 |
| AA Humanity's Last Exam | 5.14 | Accuracy (%) | 32.7 |
| Chatbot Arena (Text - English) | 1272 | Arena Score | 24.7 |
| Chatbot Arena (Text - Creative Writing) | 1211 | Arena Score | 24.2 |
Interactive version: theaggregate.ai/model?slug=jamba-1-5-mini · How It Works · Data refreshed daily, snapshot 2026-09-05.