Side-by-side comparison of Jamba 1.5 Large (AI21 Labs · Israel) and Kimi K2 Instruct (Moonshot AI · China) for self-hosted deployment of the open-weight model. Jamba 1.5 Large is rated conditional; Kimi K2 Instruct is conditional. They part ways on licence: Jamba 1.5 Large is "Jamba Open", Kimi K2 Instruct is "Modified MIT".
| Field | ||
|---|---|---|
| Summary | ||
| Verdict | Conditional SSM-Transformer hybrid (Mamba) with 256K context. Jamba Open Model License permits commercial use below $50M annual revenue; above that, paid licence required. Israel jurisdiction; EU adequacy decision in place. | Conditional 1T-parameter MoE (32B active) tuned for agentic and tool-use workflows. Modified MIT permits commercial use. Same Chinese-origin alignment and supply-chain considerations as DeepSeek and Qwen. |
| Last reviewed | 2026-04-15 | 2026-04-16 |
| Open-weight | ||
| Licence | Jamba Open | Modified MIT |
| Commercial use | Under $50M rev. | Yes |
| Training data | Undisclosed | Undisclosed |
| Origin | Israel | China |
| Performance & pricing? | ||
| Quality index | 11/100 | 26/100 |
| Speed | — | 34 tok/s |
| Blended price | $3.50/M | $1.04/M |
| Context window | — | — |
| Evidence | ||
| Sources | ||
No overlapping sources between the two entries.