whatmodel.sh / Gemma 7B (base)

Gemma 7B (base)

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

Gemma 7B (base) is an open-weights language model from Google, released February 2024. Its published size is 8.54B parameters and a 8k-token context window. Licence: gemma.

Catalogue identity notes

Maker repo google/gemma-7b @ff6768d: licence tag gemma. 8,537,680,896 parameters from the Hugging Face safetensors metadata (dense, so active = total). Released 2024-02: month the maker's Hugging Face repo was created (createdAt 2024-02-08). Pretrained Gemma 1.0 7B; the catalogue's gemma-7b is the instruction-tuned Gemma 7B IT. Ollama tags gemma:7b-text-q4_K_M, gemma:7b-text-q8_0 (sizes as listed on the tags page, observed 2026-09-29).

35.2±0.5#599 of 719 LLMsfrom 37 benchmark results across 9 sources

$ how it compares

The LLMs ranked either side of Gemma 7B (base), out of 828 with a score, then more from Google.

#modelscorewhereRAMreleased
692Qwen2.5 1.5B Instruct Alibaba36±1local~2.4GB2024-09
693MPT-30B-Chat MosaicML35±2comparison only2023-06
694Gemma 7B (base) Google35±1local~7.1GB2024-02
695Tulu 2 DPO 7B Allen Institute for AI35±1local~5.8GB2023-11
696Eurus-2-7B-SFT PRIME-RL35±3local~6.4GB2024-12
more from Google
9Gemini 4 Argon (limited preview) Google82±1limited preview2026-09
23Gemini 3.8 Flash Google77±1api $1.50/M2026-09
32Gemini 3.7 Flash Google76±1api $1.50/M2026-08

see all LLMs · find the ones that fit your machine

$ what it's good at

coding▁▁▁▃▆██▇▆▄▃▁ #416 of 458 · 71% of its score · 7 tests

Beats DeepSeek R1 Distill Qwen 7B, Qwen3.5 2B, DeepSeek R1 Distill Llama 8B here, though they rank higher overall. Trails Llama 3.2 1B Instruct, SmolLM2 1.7B Instruct, Granite 3B Code Base (2K) here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
HumanEval +335.4% ✎▁·▁▁▃▄▆▅█▃▁▁ beats 20/99easy · avg 79%▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 41%sharp · at its level · several runs
HumanEval+28.7% ✎▁·▁▁▁▃▄▄▄█▃▁ beats 17/99easy · avg 74%▇▇▇▇▇▇ 16%sharp · at its level
HumanEvalSynthesize (HumanEvalPack, 6-language average)34.4% ✎▁·▂▁▂▆▅█▄▄▁▂ beats 12/35medium · avg 64%▇▇ 4.2%at its level
MBPP+43.4% ✎▁···▁▁▃▅▇█▆▂ beats 9/75easy · avg 68%▇ 3.1%at its level
MBPP lab +144.4% ✎▁▁▁▄▅█▄▅▃▁▂▁ beats 21/81easy · avg 74%▇ 2.6%at its level · several runs
HumanEvalExplain (HumanEvalPack, 6-language average)15.0% ✎▄▇█▃▄▇▄▁▇▃·▁ beats 9/35medium · avg 49%▇ 2.3%typical
HumanEvalFix (HumanEvalPack, 6-language average)7.4% ✎▁·▁▁▄██▄█▁▅▅ beats 3/35medium · avg 56%▇ 2.0%sharp
math▁▂▃▅▆▅▄█▄▁▁▁ #291 of 398 · 4.5% of its score · 2 tests

Beats GPT-4 (0613), Athene-70B, Gemma 2 27B IT here, though they rank higher overall.

testresultvs the others in its rundifficultycountswhy
GSM8K +153.3% ✎▄··▄···▄··█▄ beats 4/5easy · avg 79%▇ 2.6%sharp · at its level · small run · several runs
MATH +124.1% ✎▃·▃····▃···█ beats 5/5medium · avg 50%▇ 1.9%small run · several runs
other tests22 not in a task · 25% of its score
testresultvs the others in its rundifficultycountswhy
EvoEval (overall)23.0% ✎▂▂▃▄▄█▁▄▅▃▁▃ beats 7/42medium · avg 55%▇▇ 5.8%typical
MedQA (USMLE, 4 options)48.7% ✎▁▁··▄▁▁··▄▃█ beats 10/17easy · avg 76%▇ 2.6%sharp · at its level
BIG-Bench Hard lab55.1% ✎▁▁▁▁▂▄▅▅█▃▄▁ beats 58/129easy · avg 82%▇ 1.9%sharp · at its level · small run
ARC Challenge lab53.2% ✎▆▁▁▂▂▅▇█▄▃▃▂ beats 62/97easy · avg 83%▇ 1.8%sharp · at its level · small run
MedMCQA49.3% ✎▂▂··▄▄▂▂·▄██ beats 12/17medium · avg 68%▇ 1.4%at its level
OpenBookQA lab52.8% ✎█▁▁▁▄▃▃▇▂▁▄▂ beats 49/65easy · avg 74%▇ 1.1%at its level · small run
WinoGrande lab72.3% ✎▁···▂▁▁▅█▇▃▁ beats 69/87easy · avg 87%▇ 1.1%at its level · small run
Natural Questions (closed book) lab23.0% ✎▁·▁·▁▂▁▂▁▄▃█ beats 19/33medium · avg 53%▇ 1.1%small run
14 more tests
testresultvs the others in its rundifficultycountswhy
MMLU lab64.3% ✎▁▁▁▁▁▂▃▅█▃▆▃ beats 82/183easy · avg 81%▇ 1.0%at its level · small run
BoolQ lab83.2% ✎▁▄▅▆▂▆▄▄█▃▄▄ beats 47/58easy · avg 92%▇ 1.0%at its level · small run
TriviaQA lab63.4% ✎▁·▁▁▄▁▂▂▅▁▃█ beats 27/38very easy · avg 85%▇ 0.9%at its level · small run
HellaSwag lab81.2% ✎▁▁▁▁▂▂▃▇█▆▅▃ beats 69/85very easy · avg 90%▇ 0.8%at its level · small run
AGIEval English (Gemma report) lab41.7% ✎▄··········█ beats 2/2easy · avg 65%▇ 0.7%at its level · small run
CommonsenseQA lab71.3% ✎▄·▄▄·▄····▄█ beats 6/6easy · avg 84%▇ 0.6%at its level · small run
ARC Easy lab81.5% ✎▁▂▃▂▃▅▇█▇▆▄▆ beats 63/65very easy · avg 90%▇ 0.5%at its level · small run
Open PT-LLM Leaderboard average (9 Portuguese tasks)64.2% ✎▃▁▁▁▁▁▁▁▂█▅▁ beats 109/276medium · avg 73%▇ 0.5%at its level · noisy
Social IQA lab51.8% ✎▂·▂██▅▅▁▂▃▆▁ beats 34/51medium · avg 65%▇ 0.5%blunt · at its level · small run
EuroEval Life in the UK (English knowledge MC), test split, 5-shot, accuracy84.9% ✎▁▁▁▁▁▁▁▂▄█▄▃ beats 98/146easy · avg 83%▇ 0.4%at its level · noisy
PIQA lab81.2% ✎▁▁▁▃▃▄▇█▄▃▁▁ beats 71/83easy · avg 87%▇ 0.3%blunt · at its level · small run
EuroEval HellaSwag (English), test split, 5-shot generative MC, accuracy33.5% ✎▃·▁▁▃▁▂▅█▇▄▁ beats 17/100medium · avg 68%▇ 0.3%noisy
TruthfulQA MC244.9% ✎▅▃█▆▁▂▅···▁▁ beats 19/31medium · avg 64%▇ 0.2%blunt · at its level
PubMedQA75.6% ✎▂·▂▆·▄▂▄▄▂█▂ beats 16/17medium · avg 77%▇ 0.2%blunt · at its level

$ where to run it

llama.cpp · recommended

version

  1. 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 ↗

  2. Run Gemma 7B (base) · needs ~5.3 GB · from huggingface.co/MaziyarPanahi
    chat in the terminal
    llama cli -hf MaziyarPanahi/gemma-7b-GGUF:Q4_K_M
  3. Or start a local server and open the address it prints (http://localhost:9931)
    llama serve -hf MaziyarPanahi/gemma-7b-GGUF:Q4_K_M

LM Studio

version

  1. Install LM Studio · Open the app once before using the lms command.

    download it (macOS / Windows / Linux) ↗

  2. Run Gemma 7B (base) · needs ~5.3 GB · from huggingface.co/MaziyarPanahi

    open in LM Studio ↗

  3. Or download it in the terminal
    lms get https://huggingface.co/MaziyarPanahi/gemma-7b-GGUF@Q4_K_M -y
  4. Then run lms ls to see its name, and lms chat followed by that name

Ollama

version

  1. Install Ollama

    download it (macOS / Windows) ↗

    Linux
    curl -fsSL https://ollama.com/install.sh | sh
  2. Run Gemma 7B (base) · needs ~5.5 GB · from ollama.com
    chat in the terminal
    ollama run gemma:7b-text-q4_K_M

Official weights and documentation: huggingface.co/google ↗

$ how the score was worked out · EvalPlus / EvalPerf published… 42%, Gemma report — original evalu… 19%, Top Leaderboard Ranking = Top… 18% and 6 more

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
EvalPlus / EvalPerf published evaluation runs3▇▇▇▇▇▇▇▇▇▇▇▇▇ 42%Apache-2.0
Gemma report — original evaluations · lab-reported18▇▇▇▇▇▇ 19%CC BY 4.0
Top Leaderboard Ranking = Top Coding Proficiency, Always? EvoEval: Evolving Coding Benchmarks via LLM — authors' own runs2▇▇▇▇▇ 18%CC BY 4.0
Granite Code Models: A Family of Open Foundation Models for Code Intelligence — authors' own runs5▇▇▇ 10%CC BY 4.0
CodeGemma: Open Code Models Based on Gemma — authors' own runs · lab-reported2▇▇ 5%CC BY 4.0
Open Medical-LLM Leaderboard results (HF dataset openlifescienceai/results)3▇ 4%Apache-2.0
EuroEval leaderboards (results collection), English generative leaderboard2▇ 1%MIT
Open Portuguese LLM Leaderboard (archived)1▇ 0%Apache-2.0
Hallucinations Leaderboard results (HF dataset hallucinations-leaderboard/results)1▇ 0%Apache-2.0
prior · pull toward the middle▇ 0%
weighted average= 35.2

$ 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