# LFM3-VL-2B ![Organization](https://img.shields.io/badge/Org-Liquid%20AI-6f42c1) ![Context](https://img.shields.io/badge/Context-128K%20tokens-1f6feb) ![License](https://img.shields.io/badge/License-lfm1.0-2da44e) ![Modality](https://img.shields.io/badge/Modality-Vision%2BText-8957e5) ![Released](https://img.shields.io/badge/Released-Oct%202026-555) > **Quick answer:** LFM3-VL-2B is Liquid AI's 1.92B-parameter on-device vision-language model (part of the LFM3 family released Oct 27, 2026), combining the LFM3-1.7B language backbone with a new in-house vision encoder. It is the first LFM vision model to accept video input and the first to support a "thinking" mode for harder visual questions. Across 62 vision benchmarks it scores an average of 71.4, matching Gemma 4 12B and within 2 points of GPT-5.6 Luna despite being far smaller, with a 128K-token context window. ## At a Glance **Where LFM3-VL-2B leads** - At least 8 points ahead of all similarly sized models across 62 vision benchmarks (71.4 average) - Video understanding (new): LVBench, MVBench, Video-MME, MotionBench - Strong GUI/screen grounding: 94.6% ScreenSpot-v2 Desktop, 55.5% ScreenSpot-Pro **Where it lags** - Math & logical reasoning (68.8 category average) trails larger Gemma 4 12B (79.3) and GPT-5.6 Luna (77.3) - Not recommended by Liquid AI for coding, multi-turn conversation, or complex reasoning - Agentic category (42.5) trails frontier-scale comparison models (GPT-5.6 Luna 56.1, Gemini 3.5 Flash Lite 62.4) **Best for:** on-device visual understanding — video Q&A, document/chart reading, spatial/embodied reasoning for robotics, and GUI grounding — where local inference and a small footprint matter more than frontier knowledge depth. ## What LFM3-VL-2B Is LFM3-VL-2B pairs the LFM3 language model (24 layers: 13 gated short convolutions, 5 sliding-window attention, 6 global attention) with a new 14-layer in-house vision encoder (gated depthwise 2D convolutions plus bidirectional attention, 16×16 patches with 2D rotary embeddings). At 1.92B total parameters (1.73B language + 164M vision + 25M projector), it targets edge visual tasks: reading documents, charts, and app screens; finding objects in images; reasoning about spatial relationships for robotics; and — new in this release — understanding video, sampled at 2 fps with 4 to 768 frames. The model also introduces reasoning: it can think step by step before answering, which Liquid reports helps most on harder visual questions such as math, charts, and multi-step problems. Liquid explicitly does not recommend it for coding, multi-turn conversation, or complex reasoning given its size. ## Specifications | Field | Value | |---|---| | Organization | Liquid AI | | Parameters | 1.92B total (1.73B LM + 164M vision encoder + 25M projector) | | Context window | 128,000 tokens (input + output) | | Architecture | LFM3 backbone (tri-hybrid) + new in-house vision encoder; thinking mode supported | | License | LFM Open License v1.0 (lfm1.0) | | Release date | October 27, 2026 | | Modality | Vision + text + video (16 languages) | ## Pricing Open weights, self-hostable — no per-token API pricing. Available on [Hugging Face](https://huggingface.co/LiquidAI/LFM3-VL-2B) in BF16, GGUF (llama.cpp/LM Studio), MLX (4-bit, Apple Silicon), and ONNX formats. ## Public Benchmark Scores | Benchmark | Score | Source | Date | |---|---|---|---| | [MME](/benchmarks/mme) | 86.9 | [LFM3-VL-2B model card](https://huggingface.co/LiquidAI/LFM3-VL-2B) | 2026-10 | | [MMStar](/benchmarks/mmstar) | 74.4 | [LFM3-VL-2B model card](https://huggingface.co/LiquidAI/LFM3-VL-2B) | 2026-10 | | [MMBench (dev EN v1.1)](/benchmarks/mmbench) | 86.1 | [LFM3-VL-2B model card](https://huggingface.co/LiquidAI/LFM3-VL-2B) | 2026-10 | | [MMMU (val)](/benchmarks/mmmu) | 55.7 | [LFM3-VL-2B model card](https://huggingface.co/LiquidAI/LFM3-VL-2B) | 2026-10 | | [MMMU-Pro](/benchmarks/mmmu-pro) | 45.7 | [LFM3-VL-2B model card](https://huggingface.co/LiquidAI/LFM3-VL-2B) | 2026-10 | | [HallusionBench](/benchmarks/hallusionbench) | 65.6 | [LFM3-VL-2B model card](https://huggingface.co/LiquidAI/LFM3-VL-2B) | 2026-10 | | [DocVQA (val)](/benchmarks/docvqa) | 96.7 | [LFM3-VL-2B model card](https://huggingface.co/LiquidAI/LFM3-VL-2B) | 2026-10 | | [ChartQA (test)](/benchmarks/chartqa) | 92.7 | [LFM3-VL-2B model card](https://huggingface.co/LiquidAI/LFM3-VL-2B) | 2026-10 | | [MathVista (mini)](/benchmarks/mathvista) | 84.2 | [LFM3-VL-2B model card](https://huggingface.co/LiquidAI/LFM3-VL-2B) | 2026-10 | | [RefCOCO-avg](/benchmarks/kazemzadeh-et-al/refcoco) | 90.5 | [LFM3-VL-2B model card](https://huggingface.co/LiquidAI/LFM3-VL-2B) | 2026-10 | | [Video-MME (v1, no subtitles)](/benchmarks/video-mme/video-mme) | 62.8 | [LFM3-VL-2B model card](https://huggingface.co/LiquidAI/LFM3-VL-2B) | 2026-10 | | [ScreenSpot-v2 (desktop)](/benchmarks/cheng-et-al/screenspot-v2) | 94.6 | [LFM3-VL-2B model card](https://huggingface.co/LiquidAI/LFM3-VL-2B) | 2026-10 | *Scores above are reported by Liquid AI (evaluated with vLLM 0.27.1, reasoning mode on) and shown for context. They are not Benchgen measurements.* **Excluded from the table above (bucket-C, documented per skill procedure):** - **MM-ToolSandbox (partial credit)** (42.5) — the model card's multimodal variant of ToolSandbox; our local `toolsandbox` entity is Apple's text-only benchmark, and the methodologies aren't confirmed equivalent — excluded rather than silently mapped, per standing never-narrow/never-widen-scope policy. - **LogicVista** (60.4) — no local Benchgen benchmark entity exists yet; candidate for a future bucket-B review. - The remaining ~48 of the model card's 62 reported vision benchmarks (CountBenchQA, POPE, BLINK, MuirBench, MMVU, MVBench, MMMB, InfographicVQA, CharXiv, OCRBench v1/v2, ScreenSpot-Pro/Mobile/Web, etc.) are omitted from this table for scope but map cleanly to existing local entities — left for a future top-up pass rather than a single oversized sync batch this cycle. ## Related Models & Benchmarks - Compare: [LFM2.5-VL-3B](/models/liquid-ai/lfm2-5-vl-3b) · [LFM3-1.7B](/models/liquid-ai/lfm3-1-7b) - Benchmarks: [MMBench](/benchmarks/mmbench) · [DocVQA](/benchmarks/docvqa) · [Video-MME](/benchmarks/video-mme/video-mme) ## Frequently Asked Questions

What is LFM3-VL-2B?

LFM3-VL-2B is Liquid AI's 1.92B-parameter on-device vision-language model, the first in the LFM3 family to support video input and a step-by-step thinking mode for harder visual questions.

What is LFM3-VL-2B's context window?

128,000 tokens for combined input and output.

Is LFM3-VL-2B open source?

Yes — released under Liquid AI's LFM Open License v1.0 (lfm1.0).

What is LFM3-VL-2B best used for?

On-device visual tasks: video Q&A, document/chart/screen reading, object finding, and spatial reasoning for robotics. Not recommended for coding, multi-turn conversation, or complex reasoning.

--- *Specs and scores sourced from Liquid AI's official Hugging Face model card and release blog post (liquid.ai/blog/lfm3). Last updated 2026-10-31.*

LFM3-VL-2B — Training jobs

Training jobs