whatmodel.sh / Gemma 7B IT

Gemma 7B IT

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

Gemma 7B IT 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

Arena entry 'gemma-7b-it' (text). LMArena's leaderboard table (arena-leaderboard space @6f2eeeb) labels 'gemma-7b-it' as 'Gemma-7B-it' (Google, Gemma license), linking https://huggingface.co/google/gemma-7b-it. Maker repo google/gemma-7b-it @9c5798d: licence tag gemma. 8,537,680,896 parameters from the Hugging Face safetensors metadata (dense, so active = total). Released 2024-02: Google Gemma releases page (ai.google.dev/gemma/docs/releases): 'February 21, 2024 Initial release of Gemma in 2B and 7B sizes'. The ollama gemma library was updated to v1.1; the tags used here are the separate original (non-v1.1) builds, which have different digests from the -v1.1- tags.

32.1±0.6#625 of 719 LLMsfrom 17 benchmark results across 11 sources

$ how it compares

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

#modelscorewhereRAMreleased
720DeepSeek-Coder 1.3B Base DeepSeek32±1local~2.4GB2023-10
721Orca 2 13B Microsoft32±2local~9.8GB2023-11
722Gemma 7B IT Google32±1local~7.1GB2024-02
723Mistral 7B Instruct v0.1 Mistral32±1local~6.1GB2023-09
724Qwen1.5 4B Chat Alibaba32±2local~4.1GB2024-02
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▁▁▁▃▆██▇▆▄▃▁ #426 of 458 · 74% of its score · 6 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, Mistral 7B Instruct v0.1 here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
HumanEval +128.7% ✎▁·▁▁▃▄▆▅█▃▁▁ beats 12/99easy · avg 79%▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 42%sharp · at its level · several runs
HumanEval+25.0% ✎▁·▁▁▁▃▄▄▄█▃▁ beats 13/99easy · avg 74%▇▇▇▇▇▇▇ 18%sharp · at its level
HumanEvalSynthesize (HumanEvalPack, 6-language average)19.8% ✎▁·▃▁▂▆▅█▄▄▁▂ beats 2/35medium · avg 64%▇▇ 5.2%at its level
MBPP+36.8% ✎▁···▁▁▃▅▇█▆▂ beats 5/75easy · avg 68%▇▇ 4.1%at its level
HumanEvalExplain (HumanEvalPack, 6-language average)21.5% ✎▄▇█▃▄▇▄▁▇▃·▁ beats 17/35medium · avg 49%▇ 2.6%typical
HumanEvalFix (HumanEvalPack, 6-language average)13.7% ✎▁·▁▁▄██▄█▁▅▅ beats 12/35medium · avg 56%▇ 2.0%sharp
chat & writing▁▁▁▃▅▄▅█▆▆▇▃ #364 of 393 · 13% of its score · 4 tests

Beats Gemma 4 E2B, Code Llama 70B Instruct, Phi-2 (2.7B) here, though they rank higher overall. Trails Gemma 3 270M IT, SmolLM2 360M Instruct, Granite 4.0 350M here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
AlpacaEval 2 length-controlled10.4% ✎▁▂▂▃▆▆█▆▄▄▁▁ beats 23/122medium · avg 39%▇▇ 4.6%typical
Polyrating rating (bias-aware re-fit of Chatbot Arena votes, mid-2024)1029.0 ✎▂▁▂▁▃▇█▆▇▄▂▂ beats 19/95avg 1201▇ 3.5%blunt
LMArena (text)1137.7 ✎▁▁▁▃▃▃▃▆▆▇█▃ beats 30/379avg 1268▇ 3.4%blunt
Arena-Hard (v0.1, GPT-4-1106 judge)7.5% ✎▂▃▁▅█▅▄▃▃▅▃▁ beats 3/60medium · avg 52%▇ 1.7%sharp · too hard for it
other tests6 not in a task · 13% of its score
testresultvs the others in its rundifficultycountswhy
EvoEval (overall)17.7% ✎▂▃▃▄▄█▁▄▅▃▁▃ beats 3/42medium · avg 55%▇▇▇ 6.7%typical
WildBench v2 WB-Score (task macro)6.6198 WB-Score ✎▁·▃▁·▃█▄▇▆▅▆ beats 1/52avg 45.199 WB-Score▇▇ 3.9%sharp
Open PT-LLM Leaderboard average (9 Portuguese tasks)48.0% ✎▃▁▁▁▁▁▁▃▂█▅▁ beats 64/276medium · avg 73%▇ 0.6%at its level · noisy
EuroEval Life in the UK (English knowledge MC), test split, 5-shot, accuracy64.0% ✎▁▁▁▁▁▁▁▃▄█▄▃ beats 34/146easy · avg 83%▇ 0.5%at its level · noisy
Vectara hallucination leaderboard, 2025 dataset (HHEM-2.3, factual consistency rate)85.2% ✎▁▁▃▃▄▄█▆▃▂▂▁ beats 14/140very easy · avg 94%▇ 0.4%blunt · noisy
EuroEval HellaSwag (English), test split, 5-shot generative MC, accuracy26.5% ✎▃·▃▁▂▁▂▅█▇▄▁ beats 8/100medium · avg 68%▇ 0.4%noisy

$ 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 IT · needs ~5.3 GB · from huggingface.co/MaziyarPanahi
    chat in the terminal
    llama cli -hf MaziyarPanahi/gemma-7b-it-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-it-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 IT · 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-it-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 IT · needs ~5.62 GB · from ollama.com
    chat in the terminal
    ollama run gemma:7b-instruct-q4_K_M

Official weights and documentation: huggingface.co/google ↗

$ how the score was worked out · EvalPlus / EvalPerf published… 50%, Top Leaderboard Ranking = Top… 21%, Granite Code Models: A Family… 10% and 8 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▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 50%Apache-2.0
Top Leaderboard Ranking = Top Coding Proficiency, Always? EvoEval: Evolving Coding Benchmarks via LLM — authors' own runs2▇▇▇▇▇▇ 21%CC BY 4.0
Granite Code Models: A Family of Open Foundation Models for Code Intelligence — authors' own runs3▇▇▇ 10%CC BY 4.0
AlpacaEval 1.0 and 2.0 leaderboards (tatsu-lab/alpaca_eval repository)1▇ 5%Apache-2.0
WildBench v2 leaderboard data (Ai2)1▇ 4%Apache-2.0
Polyrating: A Cost-Effective and Bias-Aware Rating System for LLM Evaluation (arXiv:2409.00696v3)1▇ 3%CC-BY-4.0
LMArena leaderboard dataset1▇ 3%CC BY 4.0
Arena-Hard-Auto v0.1 leaderboard (repository README, pre-v2)1▇ 2%Apache-2.0
EuroEval leaderboards (results collection), English generative leaderboard2▇ 1%MIT
Open Portuguese LLM Leaderboard (archived)1▇ 1%Apache-2.0
Vectara hallucination leaderboard (vectara/hallucination-leaderboard repository)1▇ 0%Apache-2.0
prior · pull toward the middle▇ 0%
weighted average= 32.1

$ 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