whatmodel.sh / Granite 4.0 H Tiny (7B-A1B)

Granite 4.0 H Tiny (7B-A1B)

+ contribute

Granite 4.0 H Tiny (7B-A1B) is an open-weights language model from IBM, released September 2025. Its published size is 6.94B parameters (1B active per token) and a 131k-token context window. Licence: apache-2.0.

Catalogue identity notes

IBM Granite-4.0-H-Tiny (ibm-granite/granite-4.0-h-tiny @791e0d3, created 2025-09-16, apache-2.0, 6,939,037,248 params; card architecture table '# Parameters 7B', '# Active parameters 1B'): hybrid Mamba-2 MoE instruct model. Lab rows from the h-small card's 'H Tiny MoE' column; independent EuroEval and Vectara rows.

Model information sources

Release date ↗ · Weights licence ↗

HF safetensors metadata of ibm-granite/granite-4.0-h-tiny @791e0d3; active parameters from the maker card architecture table.

47.7±0.9#385 of 719 LLMsfrom 24 benchmark results across 3 sources

$ how it compares

The LLMs ranked either side of Granite 4.0 H Tiny (7B-A1B), out of 828 with a score, then more from IBM.

#	model	score	where	RAM	released
460	Gemma 3n E4B Google	48±1	local	~6.2GB	2025-06
461	Hermes 2 Theta Llama-3 70B Nous Research	48±2	local	~47GB	2024-06
462	Granite 4.0 H Tiny (7B-A1B) IBM	48±1	local	~5.6GB	2025-09
463	DeepSeek-Coder 7B Instruct v1.5 DeepSeek	48±1	local	~5.9GB	2024-01
464	Qwen2 57B-A14B Instruct Alibaba	48±2	local	~35GB	2024-06
	more from IBM
192	Granite 4.2 30B IBM	63±1	api $0.28/M +local	~19GB	2026-08
239	Granite 4.2 8B IBM	60±1	api $0.11/M +local	~6.6GB	2026-08
276	Granite 4.2 3B IBM	57±1	api $0.052/M +local	~3.7GB	2026-08

see all LLMs · find the ones that fit your machine

$ what it's good at

agent tasks▄·▄········█ #3 of 4 * · 1.3% of its score · 1 test

Trails Granite 4.0 Micro 3B here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
BFCL v3 lab	57.6% ✎	▁▁▁▁▁▃▅█▆▅▄▁ beats 86/170	medium · avg 60%	▇ 1.3%	blunt · at its level

coding█···▄······▄ #2 of 4 · 33% of its score · 7 tests

test	result	vs the others in its run	difficulty	counts	why
HumanEval lab	83.0% ✎	▁▁▁▁▁▃▅▅█▃▂▁ beats 170/190	very easy · avg 86%	▇▇▇▇ 10%	sharp
HumanEval+ lab	76.0% ✎	▁·▁▁▂▃▄▄█▅▄▃ beats 98/145	easy · avg 81%	▇▇▇ 8.6%	sharp
MBPP lab	80.0% ✎	▁▁▁▄▅█▄▅▃▃▂▁ beats 75/81	easy · avg 79%	▇▇ 4.0%	typical
BigCodeBench lab	41.1% ✎	▅····▃·█▃▅█▃ beats 11/12	medium · avg 42%	▇ 3.2%	at its level
MultiPL-E lab	55.8% ✎	▃··▄·▅▅▇▄█▃▄ beats 10/31	medium · avg 58%	▇ 3.1%	at its level
MBPP+ lab	69.0% ✎	▁··▁▁▃▄█▆▄▁▁ beats 80/120	easy · avg 70%	▇ 2.7%	at its level
CRUXEval-O lab	39.6% ✎	▂▂██▅▂▁·▁·▁▁ beats 23/51	medium · avg 48%	▇ 1.8%	blunt · at its level

chat & writing▁▁▁▃▅▄▅█▆▆▇▃ #235 of 393 * · 4.8% of its score · 2 tests

Beats Mistral Small 4 (119B-A6.5B), Ministral 3 14B Reasoning 2512, Qwen2.5 Coder 32B Instruct here, though they rank higher overall. Trails Flan-T5 XXL, LFM2.5 1.2B Instruct, Gemma 3 4B here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
IFEval instruction-level strict lab	84.8% ✎	▄▄····▄▄█▄·█ beats 4/8	very easy · avg 88%	▇ 2.5%	typical
IFEval prompt-level strict lab	78.1% ✎	▁▁··▂··▂▇█▃▃ beats 7/26	easy · avg 81%	▇ 2.3%	typical

math▁▂▃▅▆▅▄█▄▁▁▁ #237 of 398 · 24% of its score · 5 tests

Beats Command A (03-2025), GPT-4o (2024-11-20), GPT-4o (2024-05-13) here, though they rank higher overall. Trails Tulu 2 DPO 70B, OpenThinker 7B, Olmo 3 7B Instruct here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
GSM8K lab	84.7% ✎	▂▂▂▅▅▆▅▇█▇▄▃ beats 101/146	very easy · avg 87%	▇▇▇ 7.0%	sharp
DeepMind Mathematics lab	49.9% ✎	▅····▅·█▃▃·▅ beats 6/10	medium · avg 55%	▇▇ 5.5%	at its level
Minerva MATH lab	69.6% ✎	▄·▂▂·██▆·▆▄▄ beats 18/21	easy · avg 70%	▇▇ 5.2%	at its level
GSM-Symbolic lab	81.1% ✎	▃▃····▃·▅▅█▃ beats 7/10	easy · avg 83%	▇▇ 4.4%	typical
MGSM, 5 languages (en, es, fr, ja, zh) lab	45.4% ✎	█·▄▄██···█·▄ beats 7/10	medium · avg 46%	▇ 1.8%	at its level

other tests9 not in a task · 36% of its score

test	result	vs the others in its run	difficulty	counts	why
BIG-Bench Hard lab	66.3% ✎	▁▁▁▁▂▄▅▅█▃▄▁ beats 82/129	easy · avg 76%	▇▇▇ 7.5%	sharp · at its level
GPQA (subset unspecified) lab	32.6% ✎	▂▁▂▂▄▃▆▃▅█▄▃ beats 29/104	hard · avg 37%	▇▇▇ 7.2%	sharp
MMLU lab	68.7% ✎	▁▁▁▁▁▂▃▅█▃▆▃ beats 107/183	easy · avg 74%	▇▇ 6.0%	at its level
MMLU-Pro lab	44.9% ✎	▁▁·▁▁▁▃▂▄▅█▁ beats 40/188	medium · avg 52%	▇▇ 4.9%	at its level
INCLUDE lab	53.1% ✎	▂····▂▂▄▄▄█▄ beats 9/36	medium · avg 58%	▇▇ 4.4%	at its level
MMMLU lab	61.9% ✎	▂▁▁▂▃▂▄▄▅██▃ beats 15/46	medium · avg 63%	▇ 3.7%	at its level
EuroEval HellaSwag (English), test split, 5-shot generative MC, accuracy	70.4% ✎	▃·▁▁▂▁▂▅█▇▄▁ beats 51/100	medium · avg 68%	▇ 1.3%	at its level · noisy
EuroEval Life in the UK (English knowledge MC), test split, 5-shot, accuracy	81.2% ✎	▁▁▁▁▁▁▁▂▄█▄▃ beats 71/146	easy · avg 83%	▇ 0.9%	at its level · noisy

1 more test

test	result	vs the others in its run	difficulty	counts	why
Vectara hallucination leaderboard, 2025 dataset (HHEM-2.3, factual consistency rate)	92.7% ✎	▁▁▂▃▄▄█▆▃▂▂▁ beats 41/140	very easy · avg 94%	▇ 0.5%	blunt · maxed out · noisy

$ where to run it

llama.cpp · recommended

Install llama.cpp · Homebrew and winget installs name the commands llama-cli and llama-server instead.
macOS / Linux
curl -LsSf https://llama.app/install.sh | sh
Windows
irm https://llama.app/install.ps1 | iex
other ways to install ↗
Run Granite 4.0 H Tiny (7B-A1B) · needs ~4.23 GB · from huggingface.co/ibm-granite
chat in the terminal
llama cli -hf ibm-granite/granite-4.0-h-tiny-GGUF:Q4_K_M
Or start a local server and open the address it prints (http://localhost:9931)
llama serve -hf ibm-granite/granite-4.0-h-tiny-GGUF:Q4_K_M

LM Studio

version

Install LM Studio · Open the app once before using the lms command.
download it (macOS / Windows / Linux) ↗
Run Granite 4.0 H Tiny (7B-A1B) · needs ~4.23 GB · from huggingface.co/ibm-granite
open in LM Studio ↗
Or download it in the terminal
lms get https://huggingface.co/ibm-granite/granite-4.0-h-tiny-GGUF@Q4_K_M -y
Then run lms ls to see its name, and lms chat followed by that name

Ollama

Install Ollama
download it (macOS / Windows) ↗
Linux
curl -fsSL https://ollama.com/install.sh | sh
Run Granite 4.0 H Tiny (7B-A1B) · needs ~4.2 GB · from ollama.com
chat in the terminal
ollama run granite4:tiny-h

MLX (Mac)

version

Install MLX (Mac) · Apple silicon Macs only, text only.
with uv
uv tool install mlx-lm
or
pip install mlx-lm
Run Granite 4.0 H Tiny (7B-A1B) · needs ~3.9 GB, Mac only · from huggingface.co/lmstudio-community
chat in the terminal
mlx_lm.chat --model lmstudio-community/granite-4.0-h-tiny-MLX-4bit
Or start an OpenAI-compatible server at http://localhost:8080/v1
mlx_lm.server --model lmstudio-community/granite-4.0-h-tiny-MLX-4bit

Weights and documentation: huggingface.co/ibm-granite ↗

$ how the score was worked out · ibm-granite/granite-4.0-h-sma… 97%, EuroEval leaderboards (result… 2%, Vectara hallucination leaderb… 1%

Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works

source	results	share of the score	licence
ibm-granite/granite-4.0-h-small model card · lab-reported	21	▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 97%	Apache-2.0
EuroEval leaderboards (results collection), English generative leaderboard	2	▇ 2%	MIT
Vectara hallucination leaderboard (vectara/hallucination-leaderboard repository)	1	▇ 1%	Apache-2.0
prior · pull toward the middle		▇ 0%
weighted average		= 47.7

$ improve this page

Update model information Add a place to run it Add a test result Propose a benchmark Correct a benchmark score Edit data on GitHub

Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.

data updated 2026-10-11 02:20 UTC · sources and licences · json