Liquid AI
What its model cards measure
Tall ticks are Liquid AI model cards that list the benchmark in their headline table. Gray dots are cards that leave it out. Rows are the benchmarks the lab listed most, at least twice each, in the order it started using them.
Model cards
| Date | Model | Document | Benchmarks |
|---|---|---|---|
| 27 Oct 2026 | LFM3-1.7B | LFM3: Liquid Foundation Models for specialized intelligence (Liquid AI blog) | 26 |
| 13 Oct 2026 | LFM2.5-24B-A2B | LFM2.5-24B-A2B (Liquid AI blog) | 16 |
| 12 Aug 2026 | LFM2.5-VL-3B | LFM2.5-VL-3B (Liquid AI blog) | 30 |
| 4 Aug 2026 | LFM2.5-2.6B | LFM2.5-2.6B (Liquid AI blog) | 12 |
| 28 May 2026 | LFM2.5-8B-A1B | LFM2.5-8B-A1B (Liquid AI blog) | 10 |
| 20 Jan 2026 | LFM2.5-1.2B-Thinking | LFM2.5-1.2B-Thinking: On-Device Reasoning Under 1GB (Liquid AI blog) | 9 |
| 5 Jan 2026 | LFM2.5-1.2B-Instruct | Introducing LFM2.5: The Next Generation of On-Device AI (Liquid AI blog) | 7 |
| 7 Oct 2025 | LFM2-8B-A1B | LFM2-8B-A1B: An Efficient On-device Mixture-of-Experts (Liquid AI blog) | 16 |
| 12 Aug 2025 | LFM2-VL-1.6B | LFM2-VL: Efficient Vision-Language Models (Liquid AI blog) | 12 |
| 10 Jul 2025 | LFM2-1.2B | Liquid Foundation Models v2: Our Second Series of Generative AI Models (Liquid AI blog) | 7 |
Every Liquid AI model
- LFM3-1.7B Oct 2026
- LFM2.5-24B-A2B Oct 2026
- LFM2.5-VL-3B Aug 2026
- LFM2.5-2.6B Aug 2026
- LFM2.5-8B-A1B May 2026
- LFM2.5-1.2B-Thinking Jan 2026
- LFM2.5-1.2B-Instruct Jan 2026
- LFM2-8B-A1B Oct 2025
- LFM2-VL-1.6B Aug 2025
- LFM2-1.2B Jul 2025