LFM3 Models by Liquid AI
Liquid AI releases · 18 in the last 12 months · this family litChangelog →
4 models2026Up to 256K ctxFrom $0.080/1M input
Last refreshed 2026-10-29. Next refresh: weekly.
Details
ResearcherLiquid AI
LicenseLFM Open License v1.0
Commercial useRestricted
Models4
Released2026
Max context256K
Capabilities
Vision2 of 4 models
Multimodal3 of 4 models
JSON / Tool use1 of 4 models
About
Liquid AI's third generation of Liquid Foundation Models: small open-weight text, vision, audio and omni models on a tri-hybrid architecture, all branched from the same LFM3-1.7B base.
Decision facts
- Best fit
- agenticaudiovision and multimodal work
- Capability starting point
- LFM3-VL-2B with 128K context and multimodal inputs
- Lowest tracked input
- LFM3-1.7B · $0.080/1M · Liquid AI
- Closest related family
- d1
Current Variants
Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.
4 in view
LFM3-1.7BCurrent
Use when the workload needs agentic, 256K context, and 1.7B parameters.
2026-10agentic256K context1.7B parameters
LFM3-VL-2BCurrent
Use when the workload needs vision, 128K context, and 1.9B parameters.
2026-10vision128K context1.9B parameters
LFM3-Audio-2BCurrent
Use when the workload needs audio, 32K context, and 2B parameters.
2026-10audio32K context2B parameters
LFM3-Omni-2BCurrent
Use when the workload needs multimodal, 32K context, and 2.2B parameters.
2026-10multimodal32K context2.2B parameters
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| LFM3-1.7B | Use when the workload needs agentic, 256K context, and 1.7B parameters. | 2026-10 | agentic256K context1.7B parameters | Current |
| LFM3-VL-2B | Use when the workload needs vision, 128K context, and 1.9B parameters. | 2026-10 | vision128K context1.9B parameters | Current |
| LFM3-Audio-2B | Use when the workload needs audio, 32K context, and 2B parameters. | 2026-10 | audio32K context2B parameters | Current |
| LFM3-Omni-2B | Use when the workload needs multimodal, 32K context, and 2.2B parameters. | 2026-10 | multimodal32K context2.2B parameters | Current |
Release Timeline
1 release group2026-10
4 current
LFM3-1.7B
Currentagentic256K context1.7B parameters
LFM3-Audio-2B
Currentaudio32K context2B parameters
LFM3-Omni-2B
Currentmultimodal32K context2.2B parameters
LFM3-VL-2B
Currentvision128K context1.9B parameters
Specifications(4 models)
| Model | Released | Context | Parameters | Vision | Multimodal | JSON / Tool use |
|---|---|---|---|---|---|---|
| LFM3-1.7B | 2026-10 | 256K | 1.73B | No | No | Yes |
| LFM3-VL-2B | 2026-10 | 128K | 1.92B | Yes | Yes | No |
| LFM3-Audio-2B | 2026-10 | 32K | 2B | No | Yes | No |
| LFM3-Omni-2B | 2026-10 | 32K | 2.17B | Yes | Yes | No |
