1.2B LiquidAI LFM2.5
Liquid AI's small hybrid text model for on-device use. Its official config has 16 layers, only 6 of which keep a growing attention cache; the rest use short convolutions. The 128,000-token limit is from the config. Sizes come from the official GGUF files. The LFM Open licence has a revenue threshold; read its terms before deployment. No per-level perplexity loss has been published.
128K
Max Context
3
Quant Variants
GGUF Q8_0
Best Quality
—
Perplexity rise (best build)
Quantization Variants #
Per-quant VRAM, quality loss, and inference speed on RTX 4090
Measured = run on this site · Estimated = not run here: calculated, or a figure this site has not verified · Community = public reports
| Format | Level | BPW | VRAM | PPL Loss | Speed | Source | Actions |
|---|---|---|---|---|---|---|---|
| GGUF | Q4_K_M | 5 | 0.8 GB | — | — | Estimated | |
| GGUF | Q5_K_M | 5.76 | 0.9 GB | — | — | Estimated | |
| GGUF | Q8_0 | 8.52 | 1.4 GB | — | — | Estimated |
Which file each size is for
A file size is not a memory figure: the VRAM column is the calculator's total at 4K context — weights plus KV cache plus a 10% activation buffer. Units differ on purpose: file sizes are decimal GB, as Hugging Face lists them; the VRAM column is GiB (1024³ bytes) with 2% added to the weights, so 7.12 GB of file counts as about 6.8 GiB there. A vision projector (mmproj), where the repo ships one, is a separate download and is in neither number.
- GGUF Q4_K_M[LiquidAI/LFM2.5-1.2B-Instruct-GGUF](https://huggingface.co/LiquidAI/LFM2.5-1.2B-Instruct-GGUF) · LFM2.5-1.2B-Instruct-Q4_K_M.gguf · 0.73 GB (730,895,168 bytes) · listing read 2026-11-14
- GGUF Q5_K_M[LiquidAI/LFM2.5-1.2B-Instruct-GGUF](https://huggingface.co/LiquidAI/LFM2.5-1.2B-Instruct-GGUF) · LFM2.5-1.2B-Instruct-Q5_K_M.gguf · 0.84 GB (843,354,944 bytes) · listing read 2026-11-14
- GGUF Q8_0[LiquidAI/LFM2.5-1.2B-Instruct-GGUF](https://huggingface.co/LiquidAI/LFM2.5-1.2B-Instruct-GGUF) · LFM2.5-1.2B-Instruct-Q8_0.gguf · 1.25 GB (1,246,253,888 bytes) · listing read 2026-11-14
Similar models #
[Compare with LFM2.5 8B-A1B](https://quantized.uk/tools/compare/?a=lfm2-5-1-2b&b=lfm2-5-8b-a1b)
[8B-A1.5BConsumer GPUMac / Apple SiliconQ4_K_M5.4 GB—perplexity rise](https://quantized.uk/quant-hub/lfm2-5-8b-a1b/)
LFM2.5 8B-A1B
LiquidAI LFM2.5
Liquid AI's hybrid mixture-of-experts text model, with 1.5B active parameters per token as stated by the publisher. All experts remain in memory. Its config has 24 layers with 6 full-attention layers, 32 experts per routed layer and 4 selected per token, with two leading dense layers. Sizes use the official GGUF bytes and the parameter total reported by Hugging Face, rather than the rounded 8.3B in the card. The LFM Open licence has a revenue threshold. No per-level perplexity loss has been published.
3BConsumer GPUMac / Apple SiliconQ4_K_M2.5 GB+3.5%perplexity rise
Llama 3.2 3B Instruct
Meta Llama 3.2
Tiny but capable. Runs on 4GB VRAM or 8GB RAM, even on phones via llama.cpp.
3BConsumer GPUMac / Apple SiliconQ4_K_M2.1 GB+3.2%perplexity rise
Qwen2.5 3B Instruct
Alibaba Qwen2.5
Tiny Qwen2.5 for edge devices. Runs on 4GB VRAM or Raspberry Pi class hardware.
1BConsumer GPUMac / Apple SiliconQ4_K_M0.9 GB+4.2%perplexity rise
Llama 3.2 1B Instruct
Meta Llama 3.2
Ultra-light Llama for mobile and embedded. Sub-2GB VRAM with Q4.
Running LFM2.5 1.2B Instruct locally #
At Q4_K_M and 4K of context, LFM2.5 1.2B Instruct needs about 0.8 GB — 0.7 GB of weights, 0.05 GB of KV cache and a 0.1 GB activation buffer. The smallest cards in this index that clear that comfortably are 8 GB ones — 11 of them, from the Mac M1 8G up — and 78 of the 78 cards here do. These are calculated figures, not measurements: the estimate stops counting a card as comfortable at 88% of its VRAM, which is roughly the room a desktop session needs.
What longer context costs #
Going from 4K to 32K adds about 0.33 GB, taking the total to 1.2 GB. That is gentler than the parameter count suggests: only 6 of 16 layers keep a KV cache that grows with context, so the cache scales at a fraction of the usual rate. The model's native window is 125K; holding all of it at Q4_K_M would need about 2 GB.
Which build to download #
This index only tracks LFM2.5 1.2B Instruct in GGUF, across 3 levels (Q4_K_M, Q5_K_M, Q8_0). No per-level perplexity sweep has been published for this model, so the quality column is empty rather than estimated — within one model, more bits per weight is the only ordering the data supports. No throughput has been measured for this model on this site's hardware. GGUF runs on llama.cpp and Ollama across NVIDIA, AMD and Apple silicon; AWQ and GPTQ target vLLM on CUDA and ROCm; EXL2 is ExLlamaV2 — CUDA only, and now archived, so local use rather than serving.
Common questions #
- How much VRAM does LFM2.5 1.2B Instruct need?
- About 0.8 GB at Q4_K_M with 4K of context and batch 1, rising to roughly 1.2 GB at 32K. That figure is weights plus KV cache plus a 10% activation buffer, calculated from the model's architecture rather than measured on a card.
- Will LFM2.5 1.2B Instruct run on a 8GB GPU?
- Yes — at Q4_K_M and 4K context it needs about 0.8 GB, which leaves 5.2 GB spare on a Mac M1 8G, or on any of the other 8 GB cards here (11 in all). That is the smallest size in this index that clears it comfortably; 78 of 78 do.
- Which quantization of LFM2.5 1.2B Instruct should I use?
- No published quality comparison exists for this model, so pick by footprint: Q4_K_M is the level most people run, and a higher bits-per-weight level is more faithful. All 3 levels in the index are Q4_K_M, Q5_K_M, Q8_0.
Where this model fits #
Sized at Q4_K_M with a 4K context window, smallest card first. Comfortable means the estimate uses at most 88% of the memory.
-
Runs comfortably on
-
Mac M1 8G8GB · +5.2Mac M3 8G8GB · +5.2RTX 40608GB · +7.2RTX 4060 Ti 8G8GB · +7.2Radeon RX 9060 XT 8G8GB · +7.2
-
Ships in
-
Try it yourself
How to actually run this #
Deployment guides for this model and this class of hardware.
This page's figures change when the model or the runtime does.Last updated 2026-11-14 RSS → /feed.xml