Primary use stated or inferred cautiously from official documentation.
L
LIQUID AI / LiquidAI/LFM3-VL-2B
LFM3-VL-2B
A 1.92B-parameter visual understanding model accepting text, images and video sampled at two frames per second. It produces text, not images or video.
01 / SUMMARY
The essential
Maximum input capacity when published by the source.
Documented maximum generation limit.
Verified availability channels.
Primary functional family and verified specialties.
Declared terms for API access, model weights, or self-hosting.
02 / TECHNICAL SHEET
Limits and integration
| API ID | LiquidAI/LFM3-VL-2B |
|---|---|
| Model type | Vision and documents · Multimodal |
| Access model | Open weights |
| License | LFM Open License v1.0 |
| Deployment | Self-hosted |
| Release | 27 OCT 2026 |
| Knowledge cutoff | Not published |
| Entrance | Text · Image · Video |
| Exit | Text |
| Context window | 128K tokens |
| maximum output | Not published |
| Reasoning | Not published |
| Published tools | |
| Structured Outings | Not published |
| Batch processing | Not published |
| Prompt cache | Not published |
| Fine-tuning | Not published |
| Verified platforms | Hugging Face · Transformers |
Local specialization with your own data, validating accuracy, latency, memory and usage permissions before deployment.
The cited results use reasoning mode. The card distinguishes inference modes and notes that video support in llama.cpp is still under review.
03 / CAPABILITIES
What it can do
Orientation
Local specialization with your own data, validating accuracy, latency, memory and usage permissions before deployment.
Context
128K tokens · Not published
Tools and integration
Not published
Access
Hugging Face · downloadable checkpoint
Input modalities
TextImageVideo
04 / PRICES
Documented cost
| Concept | Worth | Unit/condition |
|---|---|---|
| Standard input | No published official API rate | Standard API rate |
| Cached input | Not published | Reading reused prefixes |
| Cache write or storage | Not published | The condition varies by provider |
| Standard output | Self-hosted infrastructure cost | May include reasoning tokens |
| Batch input | Not published | Asynchronous processing |
| Batch output | Not published | Asynchronous processing |
This profile covers downloadable weights. No API rate has been verified for this checkpoint. The LFM license has its own commercial terms; it is not MIT or Apache 2.0. Budget for memory, computing and operations too.
i Prices change and may depend on level, region or context length. Check the source before making a decision.
05 / EVALUATIONS
How to read the results
“
Published results belong to the stated checkpoint and protocol. They are not transferred between base, assistant, vision, audio and decision models. Our own evaluation remains pending.
| Benchmark | Result | Metric | Source |
|---|---|---|---|
| MMStar | 74,4 / 100 | Score | View source ↗ |
| MMMU validation | 55,7 / 100 | Score | View source ↗ |
Inferama only highlights a “best result” when the metric, test set, configuration, and date allow for an equivalent comparison. The supplier's figures are presented as claims from its own source.
06 / VERSIONS
Chronology
LFM3-VL-2B
Version added to Inferama's verified catalog.
07 / SOURCES