Side-by-side comparison of Kimi K2 Instruct (Moonshot AI · China) and LFM2.5-8B-A1B (Liquid AI) for self-hosted deployment of the open-weight model. Kimi K2 Instruct is rated conditional; LFM2.5-8B-A1B is conditional. They part ways on licence: Kimi K2 Instruct is "Modified MIT", LFM2.5-8B-A1B is "LFM Open License 1.0".
| Field | ||
|---|---|---|
| Summary | ||
| Verdict | 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. | Conditional Per the published LFM Open License v1.0, Liquid AI's terms are Apache-2.0-derived but cap free commercial use at USD 10M annual company revenue — above that threshold deployers must negotiate a paid licence with [email protected]. The model itself is a strong 8.3B / 1.5B-active MoE built for CPU and edge inference, but the 38-trillion-token training corpus is undisclosed and Liquid AI is a US-based MIT spinoff, so EU regulated buyers should run their own DPIA and document the data-transfer story alongside any commercial-tier negotiation. |
| Last reviewed | 2026-04-16 | 2026-05-30 |
| Open-weight | ||
| Licence | Modified MIT | LFM Open License 1.0 |
| Commercial use | Yes | Free under $10M revenue |
| Training data | Undisclosed | Undisclosed |
| Origin | China | USA (Boston) |
| Performance & pricing? | ||
| Quality index | 26/100 | — |
| Speed | 34 tok/s | — |
| Blended price | $1.04/M | — |
| Context window | — | — |
| Evidence | ||
| Sources | ||
No overlapping sources between the two entries.