[whatmodel.sh](../../README.md) / Granite 4.0 H 350M
# Granite 4.0 H 350M
IBM · released 2025-10 · 0.34B params · apache-2.0
**score 28.9** ±1.0 · #672 of 732 · from 21 benchmark results across 1 source · † no independently run results verified
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 H 350M among every model on that task or test, worst to best.
task vs the others rank counts tests
agent tasks ▁▁▃●█▄▅▅▄▄▂▁ #281 of 313* 1.7% 1
coding ●··········█ #4 of 4 39% 7
chat & writing ▁▁▁▃●▄▆█▆▇▇▃ #337 of 402* 7.2% 2
math ▁▁●▅█▆▄█▄▂▁▁ #405 of 431 26% 5
agent tasks · #281 of 313 · 1.7% of its score
Beats [Magistral Small 1.0 (2506)](magistral-small-2506.md), [Llama 4 Scout (109B-A17B)](llama-4-scout.md), [Bielik 11B v2.3 Instruct](bielik-11b-v2-3-instruct.md) here, though they rank higher overall.
test result vs the others in its run difficulty counts why
BFCL v3 (maker) 43.3% ▁▁▁▁▁●▅█▆▅▄▁ beats 23/171 medium · avg 62% 1.7% blunt · at its level
coding · #4 of 4 · 39% of its score
Trails [Granite 4.0 350M](granite-4-0-350m.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
HumanEval (maker) 38.0% ▁▁▁▁▂▂●▆█▅▂▁ beats 62/205 very easy · avg 89% 12% sharp · at its level
HumanEval+ (maker) 35.0% ▁·▁▁▂●▄▄█▅▄▃ beats 24/145 very easy · avg 86% 10% sharp · at its level
MBPP (maker) 49.0% ▁▁▁▄▅●▆▅▄▁▂▁ beats 32/86 easy · avg 81% 4.9% at its level
MultiPL-E (maker) 14.3% ●··▄·▅▅▇▄█▃▄ beats 0/31 medium · avg 51% 3.4% typical
MBPP+ (maker) 44.0% ▁··▁▁●▄█▆▄▁▁ beats 12/120 easy · avg 72% 3.4% at its level
BigCodeBench (maker) 11.2% ●····▃·█▃▅█▃ beats 1/12 medium · avg 40% 3.1% typical
CRUXEval-O (maker) 25.5% ●▂██▅▂▁·▁·▁▁ beats 2/51 medium · avg 48% 1.9% blunt
chat & writing · #337 of 402 · 7.2% of its score
Beats [OpenThinker3 7B](openthinker3-7b.md), [Code Llama 70B Instruct](codellama-70b-instruct.md), [Gemma 4 E2B](gemma-4-e2b.md) here, though they rank higher overall.
test result vs the others in its run difficulty counts why
IFEval instruction-level strict (… 67.6% ▄●····▄▄█▄·█ beats 1/8 very easy · avg 88% 3.7% at its level
IFEval prompt-level strict (maker) 55.6% ▁▁··●··▂▇█▃▃ beats 3/26 easy · avg 81% 3.5% at its level
math · #405 of 431 · 26% of its score
Beats [Command R (03-2024)](command-r-03-2024.md), [Mistral 7B Instruct v0.3](mistral-7b-instruct-v0-3.md), [Gemma 2 2B IT](gemma-2-2b.md) here, though they rank higher overall. Trails [GPT-Neo 125M](gpt-neo-125m.md), [OPT-125M](opt-125m.md), [GPT-2 XL (1.5B)](gpt-2-xl.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
GSM8K (maker) 39.3% ▂▂▂▂▃▄●▇██▆▄ beats 71/206 very easy · avg 89% 12% sharp · at its level
GSM-Symbolic (maker) 33.7% ▃●····▃·▅▅█▃ beats 1/10 easy · avg 78% 4.4% at its level
DeepMind Mathematics (maker) 6.2% ●····▅·█▃▃·▅ beats 0/10 medium · avg 50% 3.8% typical
Minerva MATH (maker) 5.8% ▁·●▁▁▂▁█▄▁▃▃ beats 3/27 medium · avg 59% 3.1% typical
MGSM, 5 languages (en, es, fr, ja… 16.2% ●·▄▄██···█·▄ beats 1/10 medium · avg 46% 2.4% typical
other tests (6)
test result vs the others in its run difficulty counts why
BIG-Bench Hard (maker) 33.3% ▁▁▁▁●▄▅▆█▃▃▁ beats 20/155 easy · avg 75% 7.3% sharp
MMLU (maker) 36.2% ▁▁▁▁▁●▂▅██▆▄ beats 25/214 easy · avg 74% 7.0% typical
GPQA (subset unspecified) (maker) 26.1% ▂●▂▂▄▃▆▃▅█▃▃ beats 7/106 hard · avg 37% 3.6% sharp · too hard for it
MMLU-Pro (maker) 14.4% ▁●·▁▁▁▃▂▄▅█▁ beats 2/188 medium · avg 46% 2.9% typical
INCLUDE (maker) 27.1% ●····▂▂▄▄▄█▄ beats 0/36 medium · avg 52% 2.5% too hard for it
MMMLU (maker) 28.0% ●▁▁▂▃▂▄▄▅██▃ beats 0/46 medium · avg 56% 2.2% typical
source results share of the score
ibm-granite/granite-4.0-h-1b model card… 21 100% ▇▇▇▇▇▇▇▇▇▇▇▇
prior: pull toward the middle 0% ▇
weighted average = 28.9
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))
## 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 H 350M (Q4_K_M, needs ~0.22 GB, from [huggingface.co/ibm-granite](https://huggingface.co/ibm-granite/granite-4.0-h-350m-GGUF/tree/a864f823cce6e6048b5752e2816fe7a23987d790))
```sh
llama cli -hf ibm-granite/granite-4.0-h-350m-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-h-350m-GGUF:Q4_K_M
```
### 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 H 350M (Q4_K_M, needs ~0.22 GB, from [huggingface.co/ibm-granite](https://huggingface.co/ibm-granite/granite-4.0-h-350m-GGUF/tree/a864f823cce6e6048b5752e2816fe7a23987d790))
```sh
lms get https://huggingface.co/ibm-granite/granite-4.0-h-350m-GGUF@Q4_K_M -y
```
3. Then run lms ls to see its name, and lms chat followed by that name
Other versions: 4bit (~0.2 GB, Mac only) `lms get https://huggingface.co/mlx-community/granite-4.0-h-350m-4bit -y` · 8bit (~0.4 GB, Mac only) `lms get https://huggingface.co/mlx-community/granite-4.0-h-350m-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 H 350M (Q8_0, needs ~0.366 GB, from [ollama.com](https://ollama.com/library/granite4/tags))
```sh
ollama run granite4:350m-h
```
### 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 H 350M (4bit, needs ~0.2 GB, Mac only, from [huggingface.co/mlx-community](https://huggingface.co/mlx-community/granite-4.0-h-350m-4bit/tree/2b96c1f68e5349bcaec2a498a68cc177537e46d1))
```sh
mlx_lm.chat --model mlx-community/granite-4.0-h-350m-4bit
```
3. Or start an OpenAI-compatible server at http://localhost:8080/v1
```sh
mlx_lm.server --model mlx-community/granite-4.0-h-350m-4bit
```
Other versions: 8bit (~0.4 GB, Mac only) `mlx_lm.chat --model mlx-community/granite-4.0-h-350m-8bit`
Model repository and documentation: [huggingface.co/ibm-granite](https://huggingface.co/ibm-granite/granite-4.0-h-350m)
[interactive version](https://whatmodel.sh/model/granite-4-0-h-350m) · [JSON](https://whatmodel.sh/api?format=json&all=1)