LFM2.5 VL 450M — Hardware Requirements & GPU Compatibility
VisionFunctionsLFM2.5-VL-450M is Liquid AI's vision-language model with 450 million parameters, representing an update to LFM2-VL-450M that uses the LFM2.5-350M backbone plus an 86-million-parameter SigLIP2 vision encoder. The card notes gains in instruction following, stronger multilingual vision understanding across nine languages, bounding box prediction for object detection, and function calling when given text-only input. Images as large as 512x512 are handled natively; larger ones get tiled. Built for edge devices, this size runs on a phone, a CPU, or any modest GPU, while Liquid AI also provides a WebGPU browser demo. The card specifies a context length of 32,768 tokens. It comes under Liquid AI's LFM1.0 license, which is custom rather than a standard open-source one, so you should check its terms before commercial use. The model was published in April 2026 as an updated version of the earlier LFM2-VL-450M.
Runs great on a 24 GB GPU
Runs from 8 GB (Q4_K_M · 125K)Best on 24 GB (Q4_K_M · ~1024 tok/s · 125K)Macs from 8 GB (Q4_K_M · ~112 tok/s)
Other
Reference GPU · RTX 4090 24 GB
Estimates, ±20% · how we estimate
Run LFM2.5 VL 450M
One recommended way. Every other method, OS and quant is in the step-by-step guide.Each way opens a step-by-step guide for your OS and quant.
Recommended
ollama run hf.co/LiquidAI/LFM2.5-VL-450M-GGUF:Q4_K_MQ4_K_M fits fully on the RTX 4090 · 23.2 GB free · 16K context
Download: LFM2.5-VL-450M-Q4_K_M.gguf0.23 GBLiquidAI
Ways to runOther ways
each opens step-by-stepRun in cloud
Your own private copy · Q4_K_M · 24 GB GPU
- The easiest way to try this model
- Running in a few minutes, easy setup
- Chat right here or connect any app
$5+ bonus for new users
Runs on RTX 3060 12GB (11 GB headroom) · Q4_K_M
- Generation speed
- ~366 tok/s
- generation speed
- Cost per 1M output tokens
- $0.048
- per 1M output tokens
How Much VRAM Does LFM2.5 VL 450M Need?
Select a quantization level; the listed hardware adheres to it.
| Quantization | VRAM | +32K | On RTX 4090 (reference) | |
|---|---|---|---|---|
| 0.6 GB | 1 GB | 2-bit · K-quant — 2-bit quantization with K-quant improvements | Fits: 125K context · ~1170 tok/s | |
| 0.6 GB | 1 GB | 3-bit medium — 3-bit medium quantization | Fits: 125K context · ~1111 tok/s | |
| 0.6 GB | 1 GB | 4-bit · legacy — 4-bit legacy quantization | Fits: 125K context · ~1111 tok/s | |
| 0.6 GB | 1 GB | 4-bit medium — 4-bit medium quantization — most popular sweet spot | Fits: 125K context · ~1024 tok/s | |
| 0.7 GB | 1.1 GB | 5-bit medium — 5-bit medium quantization — good quality/size tradeoff | Fits: 125K context · ~950 tok/s | |
| 0.7 GB | 1.1 GB | 6-bit, very good — 6-bit quantization, very good quality | Fits: 125K context · ~885 tok/s | |
| 0.8 GB | 1.2 GB | 8-bit, near-lossless — 8-bit quantization, near-lossless | Fits: 125K context · ~799 tok/s | |
| 1.3 GB | 1.7 GB | 16-bit · bfloat — Brain floating point 16 — preferred for training | Fits: 125K context · ~520 tok/s |
Selected Q4_K_M
4-bit medium quantization0.8GBat 16K context
- Weights
- 0.3 GB
- KV cache
- 0.2 GB
- Overhead
- 0.3 GB
Which Hardware Can Run LFM2.5 VL 450M?
Q4_K_M · 0.6 GBLFM2.5 VL 450M (Q4_K_M) requires 0.6 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 1+ GB is recommended. Using the full 128K context window can add up to 1.8 GB, bringing total usage to 2.4 GB. 114 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti. 185 devices with unified memory can run LFM2.5 VL 450M, including Mac Studio M5 Ultra (256 GB), Mac Studio M5 Ultra (512 GB).
≤ 11 GB
≤ 11 GB tier →fits · 3.2 GB free · up to 125K ctx
68 GPUs · 1 Mac class · 3 device families
RTX 3080~772Simulate output on RTX 3080RTX 2080 Ti~626Simulate output on RTX 2080 TiRTX 3070 Ti~618Simulate output on RTX 3070 TiM3*8 GB~112Simulate output on iMac 24" M3 (8 GB)Orin Nano~104Simulate output on NVIDIA Jetson Orin Nano 8GB (Super)iPhone~75Simulate output on iPhone 1720 GB
fits · 19.2 GB free · up to 125K ctx
1 GPU
RX 7900 XT~750Simulate output on RX 7900 XT32–40 GB
fits · 31.2 GB free · up to 125K ctx
1 GPU · 3 Mac classes · 9 device families
RTX 5090~1820Simulate output on RTX 5090M5 Max*36 GB~503Simulate output on Mac Studio M5 Max (36 GB)M632 GB~186Simulate output on iMac 24" M6 (32 GB)RTX Spark N1X~306Simulate output on NVIDIA RTX Spark N1X Laptop (32 GB)Orin~208Simulate output on NVIDIA Jetson AGX Orin 32GB
By VRAM tier · typical GPUs and Macs at Q4_K_M
Alternatives to LFM2.5 VL 450M
LFM2.5 family · 15 models →Similar models
1| Model | Score | VRAM | tok/s | Why it is similar |
|---|---|---|---|---|
| Gemma 4 26B A4B IT DFlashz-lab | — | 0.8 GB | ~1111 | same size |
Specifications
- Publisher
- Liquid AI
- Family
- LFM2.5
- Parameters
- 449M
- Architecture
- Lfm2VlForConditionalGeneration
- Context Length
- 128,000 tokens
- Vocabulary Size
- 65,536
- Release Date
- 2026-04-08
- License
- Other
Frequently Asked Questions
How much VRAM does LFM2.5 VL 450M need?
LFM2.5 VL 450M requires 0.6 GB of VRAM at Q4_K_M, or 1.3 GB at BF16. Full 128K context adds up to 1.8 GB (2.4 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 449M × 4.8 bits ÷ 8 = 0.3 GB
KV Cache + Overhead ≈ 0.3 GB (at 2K context + ~0.3 GB framework)
Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.
KV Cache + Overhead ≈ 2.1 GB (at full 128K context)
VRAM usage by quantization
What's the best quantization for LFM2.5 VL 450M?
For LFM2.5 VL 450M, Q4_K_M (0.6 GB) offers the best balance of quality and VRAM usage. Q5_K_M (0.7 GB) provides better quality if you have the VRAM. The smallest option is Q2_K at 0.6 GB.
VRAM requirement by quantization
★ Recommended — best balance of quality and VRAM usage.
Can I run LFM2.5 VL 450M on a Mac?
Yes — MacBook Air 13" M3 (8 GB) and 158 other Macs can run LFM2.5 VL 450M. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 0.6 GB of usable unified memory (RAM minus macOS overhead).
Can I run LFM2.5 VL 450M locally?
Yes — LFM2.5 VL 450M can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.
How fast is LFM2.5 VL 450M?
At Q4_K_M, LFM2.5 VL 450M can reach ~7500 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~1024 tok/s. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = (bandwidth GB/s ÷ model GB) × efficiency
Example: NVIDIA B200 → 8000 ÷ 0.64 × 0.65 = ~8125 tok/s
Estimated speed at Q4_K_M (0.6 GB)
Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
What's the download size of LFM2.5 VL 450M?
At Q4_K_M, the download is about 0.27 GB. The full-precision BF16 version is 0.90 GB. The smallest option (Q2_K) is 0.19 GB.
Which GPUs can run LFM2.5 VL 450M?
114 consumer GPUs can run LFM2.5 VL 450M at Q4_K_M. Top options include AMD Radeon RX 5600 XT, AMD Radeon RX 5700 XT, AMD Radeon RX 580 8GB. 114 GPUs have plenty of headroom for comfortable inference.
Which devices can run LFM2.5 VL 450M?
185 devices with unified memory can run LFM2.5 VL 450M at Q4_K_M (0.6 GB), including AMD Ryzen 7 8845HS (Hawk Point) Laptop, AMD Ryzen AI 7 350 (Krackan Point) Laptop, AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10. Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.