[whatmodel.sh](../../README.md) / LFM2.5 1.2B Thinking
# LFM2.5 1.2B Thinking
Liquid AI · released 2026-01 · 1.17B params · lfm1.0
**score 49.6** ±1.7 · #379 of 732 · from 10 benchmark results across 2 sources
Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only.
## What it's good at
● marks LFM2.5 1.2B Thinking among every model on that task or test, worst to best.
task vs the others rank counts tests
agent tasks ▁▁▃▆●▄▅▅▄▄▂▁ #228 of 313* 3.1% 1
chat & writing ▁▁▁▃▆▄▆█●▇▇▃ #145 of 402* 8.3% 1
math ▁▁▃▅█●▄█▄▂▁▁ #215 of 431 38% 3
agent tasks · #228 of 313 · 3.1% of its score
Beats [o1 (2024-12-17)](o1-2024-12-17.md), [o3-mini](o3-mini.md), [DeepSeek R1 (Jan '25)](deepseek-r1.md) here, though they rank higher overall. Trails [SmolLM3 3B](smollm3-3b.md), [Jamba 1.5 Mini](jamba-1-5-mini.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
BFCL v3 (maker) 57.0% ▄▄·······▄█● beats 6/6 medium · avg 55% 3.1% blunt · at its level
chat & writing · #145 of 402 · 8.3% of its score
Beats [o3-mini](o3-mini.md), [INTELLECT-3](intellect-3.md), [MiniMax-M2](minimax-m2.md) here, though they rank higher overall.
test result vs the others in its run difficulty counts why
IFEval mean of four measures (mak… 88.4% █·█··█·██·█● beats 6/6 very easy · avg 86% 8.3% typical
math · #215 of 431 · 38% of its score
Beats [MiniMax-M2.1](minimax-m2-1.md), [DeepSeek V3 (Dec '24)](deepseek-v3.md), [GPT-4o (2024-08-06)](gpt-4o-2024-08-06.md) here, though they rank higher overall. Trails [Granite 8B Code Base (4K)](granite-8b-code-base-4k.md), [Mixtral 8x7B v0.1 (base)](mixtral-8x7b-v0-1.md), [Olmo 3 7B Instruct](olmo-3-7b-instruct.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
GSM8K (maker) 85.6% █····▄▄▄···● beats 6/6 very easy · avg 87% 20% sharp
AIME 2025 (maker) 31.7% ▄█··▄···▄··● beats 5/6 hard · avg 30% 13% sharp · at its level · noisy
MATH-500 (maker) 88.0% ▄··█▄·▄···▄● beats 6/6 easy · avg 77% 5.6% noisy
other tests (5)
test result vs the others in its run difficulty counts why
GPQA Diamond (maker) 37.9% █··········● beats 4/5 hard · avg 40% 17% sharp · noisy
MMLU-Pro (maker) 49.6% █···█·██·█●█ beats 5/6 medium · avg 52% 12% at its level
IFBench (aggregation unspecified)… 44.9% ▄·█·█······● beats 5/6 medium · avg 42% 10% at its level
EuroEval HellaSwag (English), tes… 38.9% ▃·▁▁▂●▂▅█▇▄▁ beats 20/100 medium · avg 67% 5.7% sharp · at its level · noisy
EuroEval Life in the UK (English … 75.2% ▁▁▁▁▁▁▁▂●█▄▃ beats 50/146 easy · avg 82% 3.7% at its level · noisy
source results share of the score
LiquidAI/LFM2.5-1.2B-Thinking model car… 8 89% ▇▇▇▇▇▇▇▇▇▇▇
EuroEval leaderboards 2 9% ▇
prior: pull toward the middle 1% ▇
weighted average = 49.6
Licences: LiquidAI/LFM2.5-1.2B-Thinking model card ([LFM Open License v1.0](../../licenses/cards-gap-NOTICE.txt)) · EuroEval leaderboards ([MIT](https://github.com/EuroEval/leaderboards/blob/1a77f1d2f8ba1ba05ab73287fb090d76bf41b2e0/LICENSE))
## How to run it
### llama.cpp (recommended)
1. Install llama.cpp. Homebrew and winget installs name the commands llama-cli and llama-server instead. [Other ways to install](https://github.com/ggml-org/llama.cpp/blob/master/docs/install.md)
```sh
# macOS / Linux
curl -LsSf https://llama.app/install.sh | sh
# Windows
irm https://llama.app/install.ps1 | iex
```
2. Run LFM2.5 1.2B Thinking (Q4_K_M, needs ~0.7 GB, from [huggingface.co/LiquidAI](https://huggingface.co/LiquidAI/LFM2.5-1.2B-Thinking-GGUF/tree/9584426effbfa44b677e925793a12188a36b44ce))
```sh
llama cli -hf LiquidAI/LFM2.5-1.2B-Thinking-GGUF:Q4_K_M
```
3. Or start a local server and open the address it prints (http://localhost:9931)
```sh
llama serve -hf LiquidAI/LFM2.5-1.2B-Thinking-GGUF:Q4_K_M
```
Other versions: Q8_0 (~1.2 GB) `llama cli -hf LiquidAI/LFM2.5-1.2B-Thinking-GGUF:Q8_0`
### LM Studio
1. Install LM Studio. Open the app once before using the lms command. [Download it (macOS / Windows / Linux)](https://lmstudio.ai/download)
2. Run LFM2.5 1.2B Thinking (Q4_K_M, needs ~0.7 GB, from [huggingface.co/LiquidAI](https://huggingface.co/LiquidAI/LFM2.5-1.2B-Thinking-GGUF/tree/9584426effbfa44b677e925793a12188a36b44ce))
```sh
lms get https://huggingface.co/LiquidAI/LFM2.5-1.2B-Thinking-GGUF@Q4_K_M -y
```
3. Then run lms ls to see its name, and lms chat followed by that name
Other versions: Q8_0 (~1.2 GB) `lms get https://huggingface.co/LiquidAI/LFM2.5-1.2B-Thinking-GGUF@Q8_0 -y` · 4bit (~0.7 GB, Mac only) `lms get https://huggingface.co/LiquidAI/LFM2.5-1.2B-Thinking-MLX-4bit -y` · 8bit (~1.2 GB, Mac only) `lms get https://huggingface.co/LiquidAI/LFM2.5-1.2B-Thinking-MLX-8bit -y`
### Ollama
1. Install Ollama. [Download it (macOS / Windows)](https://ollama.com/download)
```sh
# Linux
curl -fsSL https://ollama.com/install.sh | sh
```
2. Run LFM2.5 1.2B Thinking (Q4_K_M, needs ~1.2 GB, from [ollama.com](https://ollama.com/library/lfm2.5-thinking/tags))
```sh
ollama run lfm2.5-thinking:1.2b
```
Other versions: Q8_0 (~1.9 GB) `ollama run lfm2.5-thinking:1.2b-q8_0`
### MLX (Mac)
1. Install MLX (Mac). Apple silicon Macs only, text only.
```sh
# with uv
uv tool install mlx-lm
# or
pip install mlx-lm
```
2. Run LFM2.5 1.2B Thinking (4bit, needs ~0.7 GB, Mac only, from [huggingface.co/LiquidAI](https://huggingface.co/LiquidAI/LFM2.5-1.2B-Thinking-MLX-4bit/tree/0703c92f46e0dfdfd4fe3c989b5f55615c6c3031))
```sh
mlx_lm.chat --model LiquidAI/LFM2.5-1.2B-Thinking-MLX-4bit
```
3. Or start an OpenAI-compatible server at http://localhost:8080/v1
```sh
mlx_lm.server --model LiquidAI/LFM2.5-1.2B-Thinking-MLX-4bit
```
Other versions: 8bit (~1.2 GB, Mac only) `mlx_lm.chat --model LiquidAI/LFM2.5-1.2B-Thinking-MLX-8bit`
Model repository and documentation: [huggingface.co/LiquidAI](https://huggingface.co/LiquidAI/LFM2.5-1.2B-Thinking)
[interactive version](https://whatmodel.sh/model/lfm2-5-1-2b-thinking) · [JSON](https://whatmodel.sh/api?format=json&all=1)