When was LFM2.5-VL-3B released?

LFM2.5-VL-3B was released on August 12, 2026 by Liquid AI. This is the official LFM2.5-VL-3B release date tracked on LLM Stats.

How big is LFM2.5-VL-3B?

LFM2.5-VL-3B has 3.1 billion parameters. It was trained on 34.0 trillion tokens. It ships as an open-weight model, so you can download and run it on your own hardware.

Who created LFM2.5-VL-3B?

LFM2.5-VL-3B was created by Liquid AI.

What is the license for LFM2.5-VL-3B?

LFM2.5-VL-3B is released under the LFM Open License v1.0 license. This is an open-source / open-weight license that permits self-hosting.

Is LFM2.5-VL-3B multimodal?

Yes, LFM2.5-VL-3B is multimodal and can accept both text and images as input.

Where is the LFM2.5-VL-3B paper or technical report?

LFM2.5-VL-3B has a paper or technical report available at https://www.liquid.ai/blog/lfm2-5-vl-3b. Use that source for architecture, training, release and evaluation details.

What models should I compare LFM2.5-VL-3B against?

Common LFM2.5-VL-3B comparisons include LFM2.5-VL-3B vs Qwen3 VL 30B A3B Thinking, LFM2.5-VL-3B vs Qwen3 VL 8B Thinking, LFM2.5-VL-3B vs Nova 2 Sonic. Compare them side by side for benchmark scores, pricing, context window, latency and API availability.

LFM2.5-VL-3B API Pricing, Context Window & Benchmarks

Name: LFM2.5-VL-3B
Author: Liquid AI

LFM2.5-VL-3B: API Pricing, Context Window & Benchmarks

LFM2.5-VL-3B is a language model from Liquid AI, released in August 2026, with multimodal input.

LFM2.5-VL-3B is Liquid AI's open-weight vision-language model (~3.12B parameters; HF safetensors total 3,123,483,888) for on-device and edge image-text→text. Non-reasoning: answers directly for low latency. Builds on the LFM2.5-2.6B

LFM2.5-VL-3B benchmarks

Capability tiers

Standing within each category, adjusted for leaderboard depth.

Real tasks performance

High-confidence performance for LFM2.5-VL-3B across real-world prompt categories. Only 95% intervals at most 4 points wide are shown.

Performance by conversation depth

How LFM2.5-VL-3B holds up as conversations get longer.

Quality Tracker

LFM2.5-VL-3B Performance Across Datasets

Scores sourced from the model's scorecard, paper, or official blog posts

llm-stats.com - Mon Sep 07 2026

Notice missing or incorrect data?

LFM2.5-VL-3B model size

LFM2.5-VL-3B has 3.1 billion parameters and was trained on 34 trillion tokens. See how it compares to other models in the same parameter range.

ParametersTraining tokens

3.1B

34Ttokens

10885× tokens-to-params ratio

Small (3–10B)

3.1B

1B7B70B405B

LFM2.5-VL-3B latency

LFM2.5-VL-3B time to first token, sustained output throughput, and failed-request rate from live API traffic over the trailing 7 days.

LFM2.5-VL-3B examples

Recent arena outputs from LFM2.5-VL-3B, picked from the highest-ranked matchups.

LFM2.5-VL-3B license

LFM2.5-VL-3B is released under the LFM Open License v1.0 license, which permits commercial use, has 3.1B parameters.

License: LFM Open License v1.0; Commercial use allowed
Parameters: 3.1B

Liquid AI LFM Open License v1.0 — open weights with a commercial-use threshold (annual revenue $10M+ requires a separate commercial license; qualified non-profits exempt for non-commercial/research use).

LFM2.5-VL-3B resources

Official sources for LFM2.5-VL-3B: official launch post, model weights.

LFM2.5-VL-3B vs other models

The most-compared alternatives to LFM2.5-VL-3B are Qwen3 VL 30B A3B Thinking, Qwen3 VL 8B Thinking, Nova 2 Sonic. Open any pair side-by-side for benchmarks, pricing, context, and latency.