Jamba 1.6 Large: benchmark results
AI21 Labs' Jamba 1.6 Large, a hybrid SSM-Transformer model with 94B active of 398B total parameters and a 256K-token context window, built for enterprise function-calling and structured output. Provider: AI21 Labs. Released 2025-03-06. Access: API.
Aggregate Rating 1473 ± 23, rank #872 of 1656 rated models, from 45 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Enkrypt AI - Safety Risk | 27.41 | Risk Score | 51.4 |
| Enkrypt AI - Bias Risk | 83.72 | Risk Score | 50.2 |
| Wolfram LLM Benchmarking Project | 43.3 | Correct Functionality (%) | 48.4 |
| Enkrypt AI - Jailbreak Risk | 11.12 | Risk Score | 39.5 |
| BenchTable - Tech | 40.4 | Weighted Score (%) | 34.4 |
| AI Chess Leaderboard (Continuation) | 481 | Elo | 29.1 |
| AI for Education Pedagogy - Primary | 75.12 | Accuracy (%) | 28.5 |
| Enkrypt AI - Insecure Code Risk | 44.89 | Risk Score | 28 |
| AI for Education Pedagogy - Maths | 67.46 | Accuracy (%) | 27.3 |
| BenchTable - STEM | 30.5 | Weighted Score (%) | 26.5 |
| AI for Education Pedagogy | 69.86 | Accuracy (%) | 25.8 |
| AI for Education Pedagogy - Technology | 72.64 | Accuracy (%) | 25.2 |
Interactive version: theaggregate.ai/model?slug=jamba-1-6-large · How It Works · Data refreshed daily, snapshot 2026-10-20.