Model ID
LiquidAI/LFM2.5-VL-450M
Capabilities
LFM2.5 VL 450M
LFM2.5 VL 450M is a compact vision-language model designed for efficient multimodal understanding and on-device deployment, combining the LFM2.5-350M language model with an 86M SigLIP2 NaFlex vision encoder. It uses a 16-layer hybrid architecture with 1,024 hidden size, 16 attention heads, and 8 key-value heads, alternating short convolution and full-attention layers for efficient sequence processing. Supporting a 32K-token context window, it processes native-resolution images through 512×512 tiling and thumbnail encoding. The model is optimized for multilingual vision understanding, object detection, bounding box prediction, instruction following, and function calling.
About this model
LFM2.5 VL 450M is a compact vision-language model designed for efficient multimodal understanding and on-device deployment, combining the LFM2.5-350M language model with an 86M SigLIP2 NaFlex vision encoder. It uses a 16-layer hybrid architecture with 1,024 hidden size, 16 attention heads, and 8 key-value heads, alternating short convolution and full-attention layers for efficient sequence processing. Supporting a 32K-token context window, it processes native-resolution images through 512×512 tiling and thumbnail encoding. The model is optimized for multilingual vision understanding, object detection, bounding box prediction, instruction following, and function calling.
Model specifications and capabilities are published by the model author and reproduced here from HF Model Card (LiquidAI/LFM2.5-VL-450M). Released under LFM 1.0. Further reading: Blog
Deploy Private Inference Endpoint
Choose your hardware and inference engine to get deployment commands and performance benchmarks tailored to your infrastructure
Hardware
B200
1,440 GB
Engine
vLLM
SGLang
Deployment code
Access Restricted
Enter your email to get access to this content
Performance under load
Tokens/s (total)
Benchmarks measured by Vultr on B200 · vLLM. Throughput and latency vary with concurrency, input length and engine configuration, so treat these as a comparison baseline, not a service guarantee.
NVIDIA HGX B200
Unlock Next-Generation AI Power on Vultr Cloud GPU