[whatmodel.sh](../../README.md) / Gemma 2 2B IT
# Gemma 2 2B IT
Google · released 2024-07 · 2.61B params · gemma
**score 39.1** ±0.6 · #577 of 732 · from 29 benchmark results across 10 sources
Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only.
## What it's good at
● marks Gemma 2 2B IT among every model on that task or test, worst to best.
task vs the others rank counts tests
coding ▁▁▁▃●▇█▇▅▄▃▁ #412 of 469 15% 3
chat & writing ▁▁▁▃▆●▆█▆▇▇▃ #314 of 402* 5.4% 2
math ▁▁●▅█▆▄█▄▂▁▁ #408 of 431 20% 4
coding · #412 of 469 · 15% of its score
Beats [DeepSeek R1 Distill Qwen 7B](deepseek-r1-distill-qwen-7b.md), [Qwen3.5 2B](qwen3-5-2b.md), [DeepSeek R1 Distill Llama 8B](deepseek-r1-distill-llama-8b.md) here, though they rank higher overall. Trails [EXAONE 3.5 2.4B Instruct](exaone-3-5-2-4b.md), [Code Llama 7B Instruct](codellama-7b-instruct.md), [Qwen2.5 1.5B Instruct](qwen2-5-1-5b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
CRUXEval (ZeroEval, 800 problems) 21.5% ●▂▆▃▆█▆▃▂·▁▁ beats 0/47 medium · avg 57% 8.2% sharp
HumanEval 40.6% ▄·▄●·▄·▄···█ beats 3/7 easy · avg 78% 4.8% sharp · at its level
MBPP 34.0% ▄···▄·●▄█··█ beats 2/7 medium · avg 61% 2.3% at its level
chat & writing · #314 of 402 · 5.4% of its score
Beats [Gemma 4 E4B](gemma-4-e4b.md), [OpenThinker3 7B](openthinker3-7b.md), [Code Llama 70B Instruct](codellama-70b-instruct.md) here, though they rank higher overall. Trails [LFM2 350M](lfm2-350m.md), [Gemma 3 1B](gemma-3-1b.md), [Qwen3 0.6B](qwen3-0-6b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
LMArena (text) 1200.1 ▁▁▁▂▃●▃▆▆▇█▃ beats 69/385 avg 1260 4.0% blunt
IFEval (aggregation unspecified) 53.8% ▄····█·▄··█● beats 7/7 easy · avg 65% 1.5% at its level
math · #408 of 431 · 20% of its score
Beats [Command R (03-2024)](command-r-03-2024.md), [Mistral 7B Instruct v0.3](mistral-7b-instruct-v0-3.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 (ZeroEval, zero-shot) 51.6% ▁●·▁▄█▅▅▅███ beats 1/43 very easy · avg 87% 10% sharp · at its level
GSM8K 30.9% █●····▄··█▄▄ beats 2/7 easy · avg 80% 4.3% sharp · at its level
MATH (subset unspecified) 21.8% ▄···▄▄●▄··█▄ beats 3/7 medium · avg 49% 3.7% typical
MATH Level 5 (ZeroEval, 721 probl… 4.3% ●▁▁▆█▂▇▇▄··▂ beats 0/38 hard · avg 30% 1.6% sharp
other tests (20)
test result vs the others in its run difficulty counts why
MMLU-Redux (ZeroEval, 2,778 quest… 51.9% ▁·●▁▄▄▄▅█▄▂▁ beats 1/53 easy · avg 75% 19% at its level
BIG-Bench Hard 43.5% ▃▃···▃··●··▅ beats 3/7 medium · avg 63% 6.7% sharp
MMLU 56.8% ▄····▄·▄▄●·█ beats 4/7 easy · avg 76% 4.4% at its level
WildBench v2 WB-Score (task macro) 27.8260 WB-Score ▁·▁▁·▃●▄▇▆▅▆ beats 9/52 avg 43.019 WB-Score 4.2% sharp
GPQA (subset unspecified) 29.4% ▄·▄··▄··▄●·█ beats 5/7 hard · avg 37% 3.6% sharp · too hard for it
CPTUB average (SpeakLeash, 0-5 ju… 53.0% ▁▂▂▂●█▃▆▅▄▂▃ beats 19/95 medium · avg 65% 3.2% at its level
MMLU-Pro 27.2% ▄·····█··●▄█ beats 3/7 medium · avg 44% 2.9% typical
WinoGrande 68.9% █···▄●▄▄···█ beats 3/7 medium · avg 81% 2.8% at its level
HellaSwag 71.2% ▄·▄··●▄·▄·▄▄ beats 2/7 easy · avg 85% 1.9% at its level
Open Persian LLM Leaderboard v2 a… 44.4% ▂▄▂●█▄▃▆▃▅▂▂ beats 18/81 medium · avg 57% 1.8% blunt · at its level
ARC Challenge 57.4% ▄····█··▄▄●█ beats 5/7 medium · avg 72% 1.6% sharp · at its level
OpenBookQA 44.2% ▄▄··▄··▄·█●▄ beats 6/7 medium · avg 55% 1.6% typical
ZebraLogic (ZeroEval, 1,000 puzzl… 4.2% ▁●▃▆██▁▁·▁▁▁ beats 1/57 hard · avg 15% 1.2% sharp · too hard for it
BoolQ 84.4% ▄▄····▄▄·●·█ beats 4/7 easy · avg 88% 1.2% at its level
Open PT-LLM Leaderboard average (… 67.7% ▃▁▁▁▁▁▁▁▂●▅▁ beats 136/276 medium · avg 72% 0.9% at its level · noisy
SurgiQ surgical MCQ (zero-shot, l… 35.0% ▂···▂●·▃▂▃██ beats 2/17 medium · avg 53% 0.7% typical
EuroEval Life in the UK (English … 70.0% ▁▁▁▁▁▁▁▂●█▄▃ beats 38/146 easy · avg 82% 0.7% at its level · noisy
EuroEval HellaSwag (English), tes… 53.9% ▃·▁▁▂▁▂●█▇▄▁ beats 29/100 medium · avg 67% 0.6% sharp · at its level · noisy
Vectara hallucination leaderboard… 93.0% ▁▁▂▃▄●█▆▃▂▂▁ beats 43/140 very easy · avg 94% 0.4% blunt · noisy
Social IQA 60.8% ▃·····▃··●▃▅ beats 4/7 medium · avg 66% 0.3% blunt · at its level
source results share of the score
IBM Granite 3.0 Language Models technic… 15 44% ▇▇▇▇▇
WildEval ZeroEval published results 5 40% ▇▇▇▇▇
WildBench v2 leaderboard data 1 4% ▇
LMArena leaderboard dataset 1 4% ▇
SpeakLeash CPTUB 1 3% ▇
Open Persian LLM Leaderboard v2 1 2% ▇
EuroEval leaderboards 2 1% ▇
Open Portuguese LLM Leaderboard 1 1% ▇
SurgiQ: A Large-Scale Multi-Domain Benc… 1 1% ▇
Vectara hallucination leaderboard 1 0% ▇
prior: pull toward the middle 0% ▇
weighted average = 39.1
Licences: IBM Granite 3.0 Language Models technical report ([Apache-2.0](https://github.com/ibm-granite/granite-3.0-language-models/blob/5960b7fb3308588bf5f86348278ec9d6015ab647/LICENSE)) · WildEval ZeroEval published results ([Apache-2.0](../../licenses/zeroeval-published-results-Apache-2.0.txt)) · WildBench v2 leaderboard data ([Apache-2.0](https://github.com/allenai/WildBench/blob/d6b8dcaf377d173d031980f97c16e1a82618c03d/LICENSE)) · LMArena leaderboard dataset ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · SpeakLeash CPTUB ([MIT](../../licenses/hf-leaderboard-spaces-MIT.txt)) · Open Persian LLM Leaderboard v2 ([Apache-2.0](../../licenses/hf-leaderboard-spaces-Apache-2.0.txt)) · EuroEval leaderboards ([MIT](https://github.com/EuroEval/leaderboards/blob/1a77f1d2f8ba1ba05ab73287fb090d76bf41b2e0/LICENSE)) · Open Portuguese LLM Leaderboard ([Apache-2.0](../../licenses/hf-leaderboard-spaces-Apache-2.0.txt)) · SurgiQ: A Large-Scale Multi-Domain Benchmark for Evaluating Surgical Understanding in Large Language Models ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · Vectara hallucination leaderboard ([Apache-2.0](https://www.apache.org/licenses/LICENSE-2.0))
## 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 Gemma 2 2B IT (Q4_K_M, needs ~1.7 GB, from [huggingface.co/lmstudio-community](https://huggingface.co/lmstudio-community/gemma-2-2b-it-GGUF/tree/6aa72da804ad76c5dc862867bfba6256de9172c7))
```sh
llama cli -hf lmstudio-community/gemma-2-2b-it-GGUF:Q4_K_M
```
3. Or start a local server and open the address it prints (http://localhost:9931)
```sh
llama serve -hf lmstudio-community/gemma-2-2b-it-GGUF:Q4_K_M
```
Other versions: Q8_0 (~2.8 GB) `llama cli -hf lmstudio-community/gemma-2-2b-it-GGUF:Q8_0`
### 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 Gemma 2 2B IT (Q4_K_M, needs ~1.7 GB, from [huggingface.co/lmstudio-community](https://huggingface.co/lmstudio-community/gemma-2-2b-it-GGUF/tree/6aa72da804ad76c5dc862867bfba6256de9172c7))
```sh
lms get https://huggingface.co/lmstudio-community/gemma-2-2b-it-GGUF@Q4_K_M -y
```
3. Then run lms ls to see its name, and lms chat followed by that name
Other versions: Q8_0 (~2.8 GB) `lms get https://huggingface.co/lmstudio-community/gemma-2-2b-it-GGUF@Q8_0 -y` · 4bit (~1.5 GB, Mac only) `lms get https://huggingface.co/mlx-community/gemma-2-2b-it-4bit -y` · 8bit (~2.8 GB, Mac only) `lms get https://huggingface.co/mlx-community/gemma-2-2b-it-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 Gemma 2 2B IT (Q4_K_M, needs ~2.07 GB, from [ollama.com](https://ollama.com/library/gemma2/tags))
```sh
ollama run gemma2:2b-instruct-q4_K_M
```
Other versions: Q8_0 (~3.63 GB) `ollama run gemma2:2b-instruct-q8_0`
### 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 Gemma 2 2B IT (4bit, needs ~1.5 GB, Mac only, from [huggingface.co/mlx-community](https://huggingface.co/mlx-community/gemma-2-2b-it-4bit/tree/2c715097ff9c081a6ac1e5cd239e2ac756b5bd99))
```sh
mlx_lm.chat --model mlx-community/gemma-2-2b-it-4bit
```
3. Or start an OpenAI-compatible server at http://localhost:8080/v1
```sh
mlx_lm.server --model mlx-community/gemma-2-2b-it-4bit
```
Other versions: 8bit (~2.8 GB, Mac only) `mlx_lm.chat --model mlx-community/gemma-2-2b-it-8bit`
Model repository and documentation: [huggingface.co/google](https://huggingface.co/google/gemma-2-2b-it)
[interactive version](https://whatmodel.sh/model/gemma-2-2b) · [JSON](https://whatmodel.sh/api?format=json&all=1)