Total runs:
782.0K
Run Growth:
-390.9K
Growth Rate:
-48.21%
Sponsored by
Vmake Labs Video Enhancer
LiquidAI / LFM2.5-24B-A2B-Base
Model's Last Updated: October 13 2026
text-generation
Introduction of LFM2.5-24B-A2B-Base
Model Details of LFM2.5-24B-A2B-Base
LFM2.5-24B-A2B-Base
LFM2.5 is a new family of hybrid models designed for on-device deployment . LFM2.5-24B-A2B is the largest model in the family: a 24B-parameter mixture of experts with only 2B active parameters. It builds on LFM2-24B-A2B with a 131K context window, a vocabulary doubled to 128K tokens, and pre-training scaled up from ~17T to ~33T tokens.
Find more information about LFM2.5-24B-A2B in our blog post .
🗒️ Model Details
| Model | Parameters | Description |
|---|---|---|
| LFM2.5-24B-A2B-Base | 24B (2B active) | Pre-trained base model for fine-tuning |
| LFM2.5-24B-A2B | 24B (2B active) | Post-trained for agentic workloads |
LFM2.5-24B-A2B-Base is the pre-trained text-only checkpoint, used to create all the LFM2.5-24B-A2B variants. It has the following features:
- Total parameters : 24.0B
- Active parameters : 2.45B
- Number of layers : 40 (30 double-gated short convolution blocks + 10 GQA)
- Experts : 64 per MoE block, top-4 routing
- Training budget : 36 trillion tokens
- Vocabulary size : 128,000
- Context length : 131,072 tokens
- Languages : English, Arabic, Chinese, French, German, Italian, Japanese, Korean, Portuguese, Spanish, Vietnamese, Thai, Indonesian, Hindi, Russian, Polish
Pre-training is followed by two mid-training phases: 2T tokens at 32K context on curated data weighted toward reasoning, math, code, instruction following, and multilingual text, then 300B tokens at 131K context weighted toward long-context data. Tool calls and multi-turn trajectories appear throughout both phases to prepare the model for agentic workloads.
This pre-trained checkpoint is only recommended for tasks that require heavy fine-tuning, like language-specific (e.g., Japanese) or domain-specific (e.g., medical) assistants, training on proprietary data, or experimenting with novel post-training approaches. Other checkpoints to consider for your use case:
| Model | Description |
|---|---|
| LFM2.5-24B-A2B | Original post-trained model checkpoint in native format. Best for fine-tuning or inference with Transformers, vLLM, and SGLang. |
| LFM2.5-24B-A2B-GGUF | Quantized format for llama.cpp and compatible tools. Optimized for CPU inference and local deployment with reduced memory usage. |
| LFM2.5-24B-A2B-ONNX | ONNX Runtime format for cross-platform deployment. Enables hardware-accelerated inference across diverse environments (cloud, edge, mobile). |
| LFM2.5-24B-A2B-MLX | MLX format for Apple Silicon. Optimized for fast inference on Mac devices using the MLX framework. |
🏃 Inference
LFM2.5 is supported by many inference frameworks. See the Inference documentation for the full list.
| Name | Description | Docs | Notebook |
|---|---|---|---|
| Transformers | Simple inference with direct access to model internals. | Link |
|
| vLLM | High-throughput production deployments with GPU. | Link |
|
| llama.cpp | Cross-platform inference with CPU offloading. | Link |
|
| MLX | Apple's machine learning framework optimized for Apple Silicon. | Link | — |
| LM Studio | Desktop application for running LLMs locally. | Link | — |
| SGLang | High-throughput production deployments with GPU. | Link | - |
Quick start with Transformers (compatible with
transformers>=5.0.0
):
from transformers import AutoModelForCausalLM, AutoTokenizer, TextStreamer
model_id = "LiquidAI/LFM2.5-24B-A2B-Base"
model = AutoModelForCausalLM.from_pretrained(
model_id,
device_map="auto",
dtype="bfloat16",
# attn_implementation="flash_attention_2" <- uncomment on compatible GPU
)
tokenizer = AutoTokenizer.from_pretrained(model_id)
streamer = TextStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)
prompt = "What is C. elegans?"
input_ids = tokenizer.apply_chat_template(
[{"role": "user", "content": prompt}],
add_generation_prompt=True,
return_tensors="pt",
tokenize=True,
)["input_ids"].to(model.device)
output = model.generate(
input_ids,
max_new_tokens=512,
streamer=streamer,
)
🔧 Fine-Tuning
We recommend fine-tuning LFM2.5 for your specific use case to achieve the best results.
| Name | Description | Docs | Notebook |
|---|---|---|---|
| CPT ( Unsloth ) | Continued Pre-Training using Unsloth for text completion. | Link |
|
| CPT ( Unsloth ) | Continued Pre-Training using Unsloth for translation. | Link |
|
| SFT ( Unsloth ) | Supervised Fine-Tuning with LoRA using Unsloth. | Link |
|
| SFT ( TRL ) | Supervised Fine-Tuning with LoRA using TRL. | Link |
|
| DPO ( TRL ) | Direct Preference Optimization with LoRA using TRL. | Link |
|
| GRPO ( Unsloth ) | GRPO with LoRA using Unsloth. | Link |
|
| GRPO ( TRL ) | GRPO with LoRA using TRL. | Link |
|
📬 Contact
- Got questions or want to connect? Join our Discord community
- If you are interested in custom solutions with edge deployment, please contact our sales team .
Citation
@article{liquidAI202624BA2B,
author = {Liquid AI},
title = {LFM2.5-24B-A2B: Built for Fast On-Device Agents},
journal = {Liquid AI Blog},
year = {2026},
note = {www.liquid.ai/blog/lfm2-5-24b-a2b},
}
@article{liquidai2025lfm2,
title = {LFM2 Technical Report},
author = {Liquid AI},
journal = {arXiv preprint arXiv:2511.23404},
year = {2025}
}
Runs of LiquidAI LFM2.5-24B-A2B-Base on huggingface.co
855
Total runs
131
24-hour runs
467
3-day runs
685
7-day runs
685
30-day runs
More Information About LFM2.5-24B-A2B-Base huggingface.co Model
Url of LFM2.5-24B-A2B-Base
LFM2.5-24B-A2B-Base huggingface.co Url
Provider of LFM2.5-24B-A2B-Base huggingface.co
Other API from LiquidAI
Updated:September 23 2026
Total runs:
398.4K
Run Growth:
-181.1K
Growth Rate:
-44.14%
Total runs:
346.3K
Run Growth:
24.4K
Growth Rate:
7.08%
Total runs:
171.1K
Run Growth:
151.5K
Growth Rate:
98.17%
Total runs:
168.1K
Run Growth:
-378.7K
Growth Rate:
-205.04%
Total runs:
155.2K
Run Growth:
102.7K
Growth Rate:
71.57%
Total runs:
120.5K
Run Growth:
46.7K
Growth Rate:
43.38%
Total runs:
113.9K
Run Growth:
-102.3K
Growth Rate:
-88.81%
Total runs:
104.2K
Run Growth:
78.6K
Growth Rate:
75.40%
Total runs:
100.4K
Run Growth:
58.9K
Growth Rate:
58.70%
Total runs:
95.6K
Run Growth:
-52.0K
Growth Rate:
-51.95%
Total runs:
94.0K
Run Growth:
-38.8K
Growth Rate:
-41.26%
Total runs:
81.0K
Run Growth:
2.9K
Growth Rate:
3.61%
Total runs:
78.6K
Run Growth:
65.7K
Growth Rate:
83.43%
Total runs:
52.8K
Run Growth:
36.3K
Growth Rate:
68.79%
Total runs:
36.0K
Run Growth:
-3.0K
Growth Rate:
-8.21%
Total runs:
30.3K
Run Growth:
19.5K
Growth Rate:
64.33%
Total runs:
29.8K
Run Growth:
25.6K
Growth Rate:
86.05%
Total runs:
28.0K
Run Growth:
-5.1K
Growth Rate:
-18.24%
Total runs:
26.1K
Run Growth:
-2.1K
Growth Rate:
-8.04%
Total runs:
22.9K
Run Growth:
7.9K
Growth Rate:
34.65%
Total runs:
21.1K
Run Growth:
8.4K
Growth Rate:
69.93%
Total runs:
20.2K
Run Growth:
9.2K
Growth Rate:
47.85%
Total runs:
19.4K
Run Growth:
-3.1K
Growth Rate:
-15.95%
Total runs:
18.9K
Run Growth:
-71.4K
Growth Rate:
-314.96%
Total runs:
16.5K
Run Growth:
2.7K
Growth Rate:
16.31%
Total runs:
15.0K
Run Growth:
-24.0K
Growth Rate:
-159.48%
Total runs:
11.6K
Run Growth:
-6.1K
Growth Rate:
-52.59%
Total runs:
11.1K
Run Growth:
-28.7K
Growth Rate:
-258.65%
Total runs:
10.5K
Run Growth:
659
Growth Rate:
6.36%
Total runs:
10.3K
Run Growth:
2.6K
Growth Rate:
25.11%
Total runs:
9.5K
Run Growth:
-4.2K
Growth Rate:
-41.24%
Total runs:
9.2K
Run Growth:
8.0K
Growth Rate:
90.62%
Total runs:
8.6K
Run Growth:
3.0K
Growth Rate:
34.97%
Total runs:
7.6K
Run Growth:
3.8K
Growth Rate:
49.99%
Total runs:
6.4K
Run Growth:
4.2K
Growth Rate:
66.37%
Total runs:
6.1K
Run Growth:
-580
Growth Rate:
-9.47%
Total runs:
5.9K
Run Growth:
161
Growth Rate:
2.77%
Total runs:
5.8K
Run Growth:
59
Growth Rate:
1.06%
Total runs:
5.0K
Run Growth:
295
Growth Rate:
5.88%
Total runs:
4.1K
Run Growth:
2.7K
Growth Rate:
67.09%
Total runs:
4.1K
Run Growth:
2.3K
Growth Rate:
55.70%
Total runs:
4.0K
Run Growth:
-1.2K
Growth Rate:
-30.60%
Total runs:
3.9K
Run Growth:
2.2K
Growth Rate:
56.04%
Total runs:
3.5K
Run Growth:
555
Growth Rate:
15.89%
Total runs:
3.0K
Run Growth:
-879
Growth Rate:
-28.93%
Total runs:
2.9K
Run Growth:
232
Growth Rate:
7.89%
Total runs:
2.4K
Run Growth:
1.9K
Growth Rate:
80.03%
Total runs:
2.4K
Run Growth:
-253
Growth Rate:
-10.63%
Total runs:
2.3K
Run Growth:
-1.7K
Growth Rate:
-76.74%
Total runs:
2.2K
Run Growth:
-451
Growth Rate:
-20.47%
Total runs:
2.0K
Run Growth:
210
Growth Rate:
10.13%
Total runs:
1.9K
Run Growth:
1.5K
Growth Rate:
75.59%
Total runs:
1.8K
Run Growth:
96
Growth Rate:
5.30%
Total runs:
1.8K
Run Growth:
1.6K
Growth Rate:
93.66%
Total runs:
1.7K
Run Growth:
-3.0K
Growth Rate:
-103.18%
Total runs:
1.7K
Run Growth:
-478
Growth Rate:
-28.49%
Total runs:
1.6K
Run Growth:
-4.0K
Growth Rate:
-255.57%
Total runs:
1.5K
Run Growth:
-3.3K
Growth Rate:
-214.74%
Total runs:
1.5K
Run Growth:
268
Growth Rate:
17.91%
Total runs:
1.5K
Run Growth:
207
Growth Rate:
13.88%
Total runs:
1.4K
Run Growth:
746
Growth Rate:
53.40%
Total runs:
1.4K
Run Growth:
-5
Growth Rate:
-0.37%
Total runs:
1.3K
Run Growth:
-12
Growth Rate:
-0.98%
Total runs:
1.2K
Run Growth:
1.1K
Growth Rate:
89.01%
Total runs:
1.2K
Run Growth:
1.2K
Growth Rate:
96.99%
Total runs:
1.1K
Run Growth:
-1.6K
Growth Rate:
-145.50%
Total runs:
1.0K
Run Growth:
127
Growth Rate:
12.37%
Total runs:
1.0K
Run Growth:
204
Growth Rate:
20.22%
Total runs:
953
Run Growth:
31
Growth Rate:
3.31%
Total runs:
945
Run Growth:
411
Growth Rate:
43.49%
Total runs:
917
Run Growth:
641
Growth Rate:
69.90%
Total runs:
904
Run Growth:
434
Growth Rate:
48.01%
Total runs:
864
Run Growth:
-992
Growth Rate:
-114.81%
Total runs:
819
Run Growth:
-1.3K
Growth Rate:
-156.45%
Total runs:
732
Run Growth:
-1
Growth Rate:
-0.14%
Total runs:
721
Run Growth:
-42
Growth Rate:
-5.83%
Total runs:
681
Run Growth:
-9
Growth Rate:
-1.32%
Total runs:
678
Run Growth:
-80
Growth Rate:
-11.80%
Total runs:
617
Run Growth:
-811
Growth Rate:
-131.44%
Total runs:
611
Run Growth:
454
Growth Rate:
77.47%
Total runs:
606
Run Growth:
268
Growth Rate:
45.81%
Total runs:
518
Run Growth:
-674
Growth Rate:
-130.12%
Total runs:
516
Run Growth:
-17
Growth Rate:
-3.29%