whatmodel.sh / Granite 4.1 3B

Granite 4.1 3B

+ contribute
Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

Granite 4.1 3B is an open-weights language model from IBM, released April 2026. Its published size is 3B parameters. Licence: apache-2.0.

Catalogue identity notes

Comparison-only, proposed so the granite-4-1-8b card runs count. The pinned granite-4.1-8b card (@1504002) prints one Evaluation Results table with '3B Dense', '8B Dense' and '30B Dense' columns for the Granite 4.1 family. Weights: https://huggingface.co/ibm-granite/granite-4.1-3b (sha c0650403e44e78ec0262dab1c90914c65b196c4e, created 2026-04-06, apache-2.0, 3,402,836,480 params in safetensors metadata). Dense, so active = total.

49.2±1.0#359 of 719 LLMsfrom 21 benchmark results across 1 source

$ how it compares

The LLMs ranked either side of Granite 4.1 3B, out of 828 with a score, then more from IBM.

#modelscorewhereRAMreleased
428OpenThinker2 7B Open Thoughts49±2local~6.4GB2025-04
429Aya 23 35B Cohere For AI49±3local~22GB2024-05
430Granite 4.1 3B IBM49±1local~3.7GB2026-04
431OpenCodeInterpreter-DS-6.7B M-A-P49±1local~5.8GB2024-02
432Artigenz-Coder-DS-6.7B Artigenz49±1local~5.8GB2024-04
more from IBM
192Granite 4.2 30B IBM63±1api $0.28/M +local~19GB2026-08
239Granite 4.2 8B IBM60±1api $0.11/M +local~6.6GB2026-08
276Granite 4.2 3B IBM57±1api $0.052/M +local~3.7GB2026-08

see all LLMs · find the ones that fit your machine

$ what it's good at

agent tasks▁▁▃▆█▆▅▅▅▄▃▁ #208 of 300 * · 1.4% of its score · 1 test

Beats o1 (2024-12-17), o3-mini, DeepSeek R1 (Jan '25) here, though they rank higher overall. Trails SmolLM3 3B, Jamba 1.5 Mini here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
BFCL v3 lab60.8% ✎▁▁▁▁▁▃▅█▆▅▄▁ beats 100/170medium · avg 61%▇ 1.4%blunt · at its level · small run
coding▁▁▁▃▆██▇▆▄▃▁ #277 of 458 · 33% of its score · 7 tests

Beats Qwen3-VL 32B Instruct, Solar Open 100B (102B-A12B), GLM-4.7-Flash here, though they rank higher overall. Trails DeepSeek LLM 67B Chat, Yi 1.5 9B Chat, Mistral NeMo Instruct (2407) here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
HumanEval lab81.7% ✎▁▁▁▁▁▃▅▅█▃▂▁ beats 165/190easy · avg 80%▇▇▇▇ 10%sharp · small run
HumanEval+ lab76.8% ✎▁·▁▁▂▃▄▄█▅▄▃ beats 99/145easy · avg 76%▇▇▇ 8.4%sharp · at its level · small run
MBPP lab71.2% ✎▁▁▁▄▅█▄▅▃▁▂▁ beats 67/81easy · avg 76%▇▇ 3.9%at its level · small run
MultiPL-E lab52.5% ✎▃··▄·▅▅▇▄█▃▄ beats 9/31medium · avg 56%▇ 3.3%at its level · small run
BigCodeBench lab32.2% ✎▅····▃·█▃▅█▃ beats 6/12hard · avg 32%▇ 2.9%at its level · small run
MBPP+ lab62.2% ✎▁··▁▁▃▄█▆▄▁▁ beats 50/120easy · avg 67%▇ 2.9%at its level · small run
CRUXEval-O lab40.8% ✎▂▂██▅▂▁·▁·▁▁ beats 28/51medium · avg 48%▇ 1.8%blunt · at its level · small run
chat & writing▁▁▁▃▅▄▅█▆▆▇▃ #220 of 393 * · 2.5% of its score · 1 test

Beats Granite 4.2 8B, Mistral Small 4 (119B-A6.5B), Olmo 3.1 32B Think here, though they rank higher overall. Trails LFM2.5 1.2B Instruct, Gemma 3 4B, Olmo 3 7B Instruct here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
IFEval (aggregation unspecified) lab82.3% ✎▁▁▃▁▂▄▄▆█▃▂▁ beats 61/141easy · avg 82%▇ 2.5%small run
math▄··········█ #3 of 3 · 25% of its score · 5 tests
testresultvs the others in its rundifficultycountswhy
GSM8K lab86.9% ✎▂▂▂▅▅▆▅▇█▇▄▃ beats 110/146very easy · avg 88%▇▇▇▇ 9.1%sharp · small run
Minerva MATH lab67.9% ✎▄·▂▂·██▆·▆▄▄ beats 17/21easy · avg 68%▇▇ 5.5%at its level · small run
DeepMind Mathematics lab64.6% ✎▅····▅·█▃▃·▅ beats 8/10easy · avg 67%▇▇ 5.5%at its level · small run
GSM-Symbolic lab81.3% ✎▃▃····▃·▅▅█▃ beats 8/10easy · avg 78%▇ 3.0%small run
MGSM, 5 languages (en, es, fr, ja, zh) lab70.0% ✎█·▄▄██···█·▄ beats 8/10medium · avg 63%▇ 2.2%at its level · small run
other tests7 not in a task · 37% of its score
testresultvs the others in its rundifficultycountswhy
BIG-Bench Hard lab75.8% ✎▁▁▁▁▂▄▅▅█▃▄▁ beats 104/129easy · avg 74%▇▇▇▇ 8.9%sharp · at its level · small run
MMLU lab67.0% ✎▁▁▁▁▁▂▃▅█▃▆▃ beats 95/183medium · avg 70%▇▇ 6.1%at its level · small run
GPQA (subset unspecified) lab31.7% ✎▂▁▂▂▄▃▆▃▅█▄▃ beats 25/104hard · avg 37%▇▇ 6.1%sharp · small run
MMLU-Pro lab49.8% ✎▁▁·▁▁▁▃▃▄▅█▁ beats 49/188medium · avg 52%▇▇ 4.7%at its level · small run
INCLUDE lab52.0% ✎▂····▂▂▄▄▄█▄ beats 8/36medium · avg 55%▇▇ 4.2%at its level · small run
MMMLU lab57.6% ✎▂▁▁▂▃▂▄▄▅██▃ beats 14/46medium · avg 61%▇▇ 3.8%at its level · small run
SimpleQA lab3.7% ✎▄▄▄▆█▆███·▂█ beats 2/32very hard · avg 4.8%▇ 3.7%too hard for it · small run

$ where to run it

Ollama · recommended

  1. Install Ollama

    download it (macOS / Windows) ↗

    Linux
    curl -fsSL https://ollama.com/install.sh | sh
  2. Run Granite 4.1 3B · needs ~2.1 GB · from ollama.com
    chat in the terminal
    ollama run granite4.1:3b
$ how the score was worked out · ibm-granite/granite-4.1-8b mo… 100%

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

sourceresultsshare of the scorelicence
ibm-granite/granite-4.1-8b model card · lab-reported21▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 100%Apache-2.0
prior · pull toward the middle▇ 0%
weighted average= 49.2

$ improve this page

Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit 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 · sources and licences · json