[whatmodel.sh](../../README.md) / Granite 4.0 1B
# Granite 4.0 1B
IBM · released 2025-10 · 1.63B params · apache-2.0
**score 42.1** ±0.8 · #531 of 732 · from 32 benchmark results across 3 sources
Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only.
## What it's good at
● marks Granite 4.0 1B among every model on that task or test, worst to best.
task vs the others rank counts tests
agent tasks ▁▁▃●█▄▅▅▄▄▂▁ #245 of 313* 1.8% 1
coding █··········● #2 of 4 24% 7
chat & writing ▁▁▁▃▆▄●█▆▇▇▃ #252 of 402 6.0% 3
math ▁▁▃▅●▆▄█▄▂▁▁ #320 of 431 29% 8
agent tasks · #245 of 313 · 1.8% of its score
Beats [o3-mini](o3-mini.md), [Llama 4 Maverick (402B-A17B)](llama-4-maverick.md), [DeepSeek R1 0528 Qwen3 8B](deepseek-r1-0528-qwen3-8b.md) here, though they rank higher overall.
test result vs the others in its run difficulty counts why
BFCL v3 52.4% ▄▄·······▄●█ beats 4/6 medium · avg 62% 1.8% blunt · at its level · several runs
coding · #2 of 4 · 24% of its score
Trails [Granite 4.0 H 1B](granite-4-0-h-1b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
HumanEval (maker) 74.0% ▁▁▁▁▂▂▆▆●▅▂▁ beats 152/205 very easy · avg 89% 6.4% sharp · at its level
HumanEval+ (maker) 69.0% ▁·▁▁▂▃▄▄●▅▄▃ beats 76/145 very easy · avg 86% 5.8% sharp · at its level
BigCodeBench (maker) 30.2% ▅····▃·●▃▅█▃ beats 5/12 medium · avg 40% 2.9% at its level
MultiPL-E (maker) 32.2% ▃··●·▅▅▇▄█▃▄ beats 2/31 medium · avg 51% 2.9% at its level
MBPP (maker) 65.0% ▁▁▁▄▅█▆●▄▁▂▁ beats 60/86 easy · avg 81% 2.7% at its level
MBPP+ (maker) 57.0% ▁··▁▁▃▄●▆▄▁▁ beats 37/120 easy · avg 72% 2.1% at its level
CRUXEval-O (maker) 33.1% ▂▂●█▅▂▁·▁·▁▁ beats 11/51 medium · avg 48% 1.4% blunt · at its level
chat & writing · #252 of 402 · 6.0% of its score
Beats [Mistral Small 4 (119B-A6.5B)](mistral-small-4.md), [Qwen2.5 Coder 32B Instruct](qwen2-5-coder-32b.md), [Ministral 3 14B Reasoning 2512](ministral-3-14b-reasoning.md) here, though they rank higher overall. Trails [Flan-T5 XXL](flan-t5-xxl.md), [Granite 4.0 H 1B](granite-4-0-h-1b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
IFEval mean of four measures 79.6% █·█··█·●█·██ beats 3/6 very easy · avg 86% 2.6% typical
IFEval prompt-level strict (maker) 73.9% ▁▁··▂··●▇█▃▃ beats 4/26 easy · avg 81% 1.8% at its level
IFEval instruction-level strict (… 80.8% ▄▄····●▄█▄·█ beats 2/8 very easy · avg 88% 1.6% typical
math · #320 of 431 · 29% of its score
Beats [GPT-4 (0613)](gpt-4-0613.md), [Athene-70B](athene-70b.md), [Gemma 2 27B IT](gemma-2-27b.md) here, though they rank higher overall. Trails [StarCoder2 7B](starcoder2-7b.md), [DeepSeek R1 Distill Qwen 1.5B](deepseek-r1-distill-qwen-1-5b.md), [Tulu 2 DPO 7B](tulu-2-dpo-7b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
GSM8K 73.4% █····▄▄●···█ beats 4/6 very easy · avg 87% 14% sharp · at its level · several runs
DeepMind Mathematics (maker) 34.0% ▅····●·█▃▃·▅ beats 2/10 medium · avg 50% 3.9% at its level
Minerva MATH (maker) 45.3% ▁·▃▁▁▂▁█●▁▃▃ beats 19/27 medium · avg 59% 2.9% at its level
GSM-Symbolic (maker) 72.3% ▃▃····▃·●▅█▃ beats 3/10 easy · avg 78% 2.6% at its level
MGSM, 5 languages (en, es, fr, ja… 37.8% █·▄▄●█···█·▄ beats 4/10 medium · avg 46% 2.0% at its level
AIME 2025 3.3% ▄█··●···▄··█ beats 3/6 hard · avg 30% 1.8% sharp · noisy
MATH-500 44.8% ▄··●▄·▄···▄▄ beats 1/6 easy · avg 77% 1.6% at its level · noisy
OTIS mock AIME 2024–2025 0.8% ●▃▄▃▃▃▇▇█▃▂▅ beats 6/197 very hard · avg 7.2% 0.3% sharp · too hard for it · noisy
other tests (9)
test result vs the others in its run difficulty counts why
BIG-Bench Hard (maker) 60.4% ▁▁▁▁▂▄▅●█▃▃▁ beats 88/155 easy · avg 75% 7.5% sharp · at its level
GPQA (subset unspecified) (maker) 29.9% ▂▁▂▂●▃▆▃▅█▃▃ beats 21/106 hard · avg 37% 7.0% sharp · too hard for it
MMLU (maker) 59.4% ▁▁▁▁▁▂▂▅●█▆▄ beats 82/214 easy · avg 74% 6.4% at its level
MMLU-Pro 33.5% █···█·█●·███ beats 3/6 medium · avg 46% 5.7% several runs
GPQA Diamond 24.0% ●▁▁▁▂▄▆▄▅▇█▅ beats 1/211 hard · avg 39% 4.8% sharp · too hard for it · several runs
INCLUDE (maker) 42.1% ▂····●▂▄▄▄█▄ beats 3/36 medium · avg 52% 2.8% typical
MMMLU (maker) 45.0% ▂▁▁▂●▂▄▄▅██▃ beats 8/46 medium · avg 56% 2.3% at its level
IFBench (aggregation unspecified) 21.0% ▄·●·█······█ beats 2/6 medium · avg 42% 2.0% at its level
Chess Puzzles (Epoch) 0.0% ●▁▁▁▁▂▃▃▃▂▁▁ beats 11/137 very hard · avg 5.8% 0.1% sharp · too hard for it · noisy
source results share of the score
ibm-granite/granite-4.0-h-1b model card… 21 76% ▇▇▇▇▇▇▇▇▇
LiquidAI/LFM2.5-1.2B-Thinking model card 8 20% ▇▇
Epoch AI 3 3% ▇
prior: pull toward the middle 0% ▇
weighted average = 42.1
Licences: ibm-granite/granite-4.0-h-1b model card ([Apache-2.0](https://huggingface.co/ibm-granite/granite-4.0-h-1b/blob/d18cca4c121edb87d022116d281ce212c9136f57/README.md)) · LiquidAI/LFM2.5-1.2B-Thinking model card ([LFM Open License v1.0](../../licenses/cards-gap-NOTICE.txt)) · Epoch AI ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/))
## 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 Granite 4.0 1B (Q4_K_M, needs ~1 GB, from [huggingface.co/ibm-granite](https://huggingface.co/ibm-granite/granite-4.0-1b-GGUF/tree/b27c2fe3f211b7f44e80fa620177aea371099aaa))
```sh
llama cli -hf ibm-granite/granite-4.0-1b-GGUF:Q4_K_M
```
3. Or start a local server and open the address it prints (http://localhost:9931)
```sh
llama serve -hf ibm-granite/granite-4.0-1b-GGUF:Q4_K_M
```
Other versions: Q8_0 (~1.7 GB) `llama cli -hf ibm-granite/granite-4.0-1b-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 Granite 4.0 1B (Q4_K_M, needs ~1 GB, from [huggingface.co/ibm-granite](https://huggingface.co/ibm-granite/granite-4.0-1b-GGUF/tree/b27c2fe3f211b7f44e80fa620177aea371099aaa))
```sh
lms get https://huggingface.co/ibm-granite/granite-4.0-1b-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 (~1.7 GB) `lms get https://huggingface.co/ibm-granite/granite-4.0-1b-GGUF@Q8_0 -y` · 4bit (~0.9 GB, Mac only) `lms get https://huggingface.co/mlx-community/granite-4.0-1b-4bit -y` · 8bit (~1.7 GB, Mac only) `lms get https://huggingface.co/mlx-community/granite-4.0-1b-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 Granite 4.0 1B (BF16, needs ~4.41 GB, from [ollama.com](https://ollama.com/library/granite4/tags))
```sh
ollama run granite4:1b
```
### 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 Granite 4.0 1B (4bit, needs ~0.9 GB, Mac only, from [huggingface.co/mlx-community](https://huggingface.co/mlx-community/granite-4.0-1b-4bit/tree/29f108a4e2ff99079a173004871604c2f1e12c1f))
```sh
mlx_lm.chat --model mlx-community/granite-4.0-1b-4bit
```
3. Or start an OpenAI-compatible server at http://localhost:8080/v1
```sh
mlx_lm.server --model mlx-community/granite-4.0-1b-4bit
```
Other versions: 8bit (~1.7 GB, Mac only) `mlx_lm.chat --model mlx-community/granite-4.0-1b-8bit`
Model repository and documentation: [huggingface.co/ibm-granite](https://huggingface.co/ibm-granite/granite-4.0-1b)
[interactive version](https://whatmodel.sh/model/granite-4-0-1b) · [JSON](https://whatmodel.sh/api?format=json&all=1)