Lfm25 24b needs about 14.7 GB of memory at IQ3_S (a 11.5 GB file) with an 8K context window, so it fits on a 16 GB GPU. This repository has 5 quants, from 5.8 GB to 19.7 GB (Q6_K). Longer context needs more: at 32K tokens the KV cache alone adds about 10.7 GB.
24B-parameter reasoning model with 2B active parameters for on-device agents
Lfm25 24b is a 24.0B-parameter model with only 2.45B active parameters per token. It supports 131,072‑token context and is optimized for fast, private on‑device reasoning. The model is a mixture of experts with 64 experts per block and top‑4 routing.
Good at
- Fast on‑device reasoning
- Tool‑calling and function execution
- Large context handling
Limits
- No vision support
- Requires 8k context for memory estimates
- License is other
Best for
- Agentic workflows
- Tool‑enabled assistants
- On‑device reasoning tasks
Who it’s for. Developers building private agents that need robust reasoning and tool use on local hardware. The model is ideal for applications where inference speed and low active parameter count are critical.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 5.8 GB | 9.0 GB | Too big | Get file → | |
| IQ2_XXS | 8.9 GB | 12.1 GB | Too big | Get file → | |
| IQ3_S | 11.5 GB | 14.7 GB | Too big | Get file → | |
| IQ4_XS | 14.9 GB | 18.0 GB | Too big | Get file → | |
| Q6_K | 19.7 GB | 22.9 GB | Too big | Get file → |
For 8 GB VRAM, use IQ2_XXS (8.89 GB). For 10 GB VRAM, IQ3_S (11.53 GB) is suitable. If 16 GB VRAM is available, IQ4_XS (14.85 GB) or Q6_K (19.68 GB) can be used for higher fidelity. Estimates are approximate.
What GPU do you need for Lfm25 24b?
These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.
- 4 GB VRAM: no quant fits at an 8K context.
- 6 GB VRAM: no quant fits at an 8K context.
- 8 GB VRAM: no quant fits at an 8K context.
- 12 GB VRAM: the best quant that fits is the main file (5.8 GB file, about 9.0 GB in use).
- 16 GB VRAM: the best quant that fits is IQ3_S (11.5 GB file, about 14.7 GB in use).
- 24 GB VRAM: the best quant that fits is Q6_K (19.7 GB file, about 22.9 GB in use).
Which Lfm25 24b quant should you choose?
IQ3_S (11.5 GB) is the middle option in this repository. A reasonable choice when a 4-bit file will not fit, mostly on larger models. Quality loss is noticeable but often acceptable.
If memory is tight, the smallest file is the first one in the table at 5.8 GB, with the quality trade-offs that come with fewer bits.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Lfm25 24b
How much VRAM does Lfm25 24b need?
About 14.7 GB at IQ3_S with an 8K context window. That is the 11.5 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.
Can I run Lfm25 24b on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 9.0 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Lfm25 24b GGUF?
It depends on the quant: file is 5.8 GB, IQ2_XXS is 8.9 GB, IQ3_S is 11.5 GB, IQ4_XS is 14.9 GB, Q6_K is 19.7 GB.
What context length does Lfm25 24b support?
The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 2.7 GB at 8K and 10.7 GB at 32K.
Where can I download Lfm25 24b GGUF?
The files are on Hugging Face in the PollardWeights/LFM2.5-24B-A2B-Pollard repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does Lfm25 24b use?
Hugging Face lists the license as other. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.
Can I run Lfm25 24b on 8 GB VRAM?
Yes, the IQ2_XXS quant is 8.89 GB and fits in 8 GB VRAM with room for context.
Does this model support vision?
No, it is a text‑only reasoning model.
What context length does it support?
It supports up to 131,072 tokens.