[whatmodel.sh](../../README.md) / Gemma 3 27B
# Gemma 3 27B
Google · released 2025-03 · 27.4B params · gemma
**score 53.4** ±0.5 · #311 of 732 · from 85 benchmark results across 29 sources
Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only.
## What it's good at
● marks Gemma 3 27B among every model on that task or test, worst to best.
task vs the others rank counts tests
agent tasks ▁▁▃●█▄▅▅▄▄▂▁ #249 of 313 6.5% 7
coding ▁▁▁▃▅●█▇▅▄▃▁ #296 of 469 14% 11
chat & writing ▁▁▁▃▆▄▆█●▇▇▃ #152 of 402 7.1% 6
math ▁▁▃▅█●▄█▄▂▁▁ #204 of 431 15% 8
vision & documents ▁▁▂▂▁▃●▂▆▇█▄ #112 of 153 10% 14
agent tasks · #249 of 313 · 6.5% of its score
Beats [o3-mini](o3-mini.md), [Llama 4 Maverick (402B-A17B)](llama-4-maverick.md), [DeepSeek R1 0528 Qwen3 8B](deepseek-r1-0528-qwen3-8b.md) here, though they rank higher overall. Trails [Granite 4.0 1B](granite-4-0-1b.md), [SmolLM3 3B](smollm3-3b.md), [Qwen3-VL 2B Instruct](qwen3-vl-2b-instruct.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
Tau2 – airline (maker) 39.0% ▃·●··▃·····█ beats 1/5 medium · avg 42% 1.4% at its level
τ²-bench Telecom (maker) 3.1% ●····█··▄·▄▄ beats 0/5 very hard · avg 13% 1.2% sharp · noisy
Terminal-Bench Hard (maker) 4.0% █●··█·█·█··█ beats 1/5 very hard · avg 4.7% 1.1% sharp · too hard for it
BFCL V4 (December 2025 archive) 29.5% ▁▁▁▂█●▅▅▆▂▂▂ beats 29/77 hard · avg 31% 1.1% at its level
τ²-bench aggregate (maker) 16.2% ●·▄··▄····█▄ beats 0/5 hard · avg 34% 0.8% at its level
BFCL v3 59.1% ▁▁▁▁▁▃▅█●▄▁▁ beats 80/114 medium · avg 51% 0.5% blunt · at its level
Tau2 – retail (maker) 6.6% ●····▄··▄·▄█ beats 0/5 medium · avg 40% 0.4% at its level · noisy
coding · #296 of 469 · 14% of its score
Beats [Qwen3-VL 32B Instruct](qwen3-vl-32b-instruct.md), [Solar Open 100B (102B-A12B)](solar-open-100b.md), [Qwen3-VL 30B-A3B Instruct](qwen3-vl-30b-a3b-instruct.md) here, though they rank higher overall. Trails [DeepSeek LLM 67B Chat](deepseek-llm-67b-chat.md), [WizardLM 70B V1.0](wizardlm-70b-v1-0.md), [Yi 1.5 9B Chat](yi-1-5-9b-chat.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
Aider polyglot (225 exercises, tw… 4.9% ●·▂▃▅▂▃█▆▄▂▂ beats 1/49 very hard · avg 7.6% 3.1% sharp
HumanEval (maker) 87.8% ▁▁▁▁▂▂▆▆█▅●▁ beats 194/205 easy · avg 83% 1.7% sharp
Luau Leetcode fully solved (Torpe… 16.3% ▁▁●▁·▁▂▃▅▃█▇ beats 3/49 hard · avg 17% 1.6% sharp
WebApp1K-Duo-React pass@1 (onekq-… 19.8% ▁··●·▁▁▁▆█▇█ beats 1/45 hard · avg 29% 1.4% at its level
VisEditBench strict pass rate (1,… 43.0% ▄··█·●··▄█·▄ beats 3/7 hard · avg 33% 1.2% at its level
LiveCodeBench (version unspecifie… 29.7% ▁···▁▁●█▇▃▄▁ beats 5/38 hard · avg 23% 1.1% at its level
WebApp1K-React pass@1 (onekq-ai) 65.5% ▃▁▁█▁▄●▃▄▇▁▄ beats 15/31 medium · avg 40% 1.1% sharp · at its level · noisy
MBPP (maker) 74.4% ▁▁▁▄▅█▆▅●▁▂▁ beats 78/86 easy · avg 70% 1.1% typical
SciCode (maker) 21.0% ●▄·······▄▄▄ beats 0/5 hard · avg 21% 0.8% typical
LiveCodeBench v6 (Feb–May 2025) (… 29.1% ●··▄·▄···▄·█ beats 0/5 medium · avg 43% 0.8% at its level
BIRD-SQL (dev) (maker) 54.4% ▂·······▂▂█● beats 8/11 medium · avg 43% 0.4% at its level
chat & writing · #152 of 402 · 7.1% of its score
Beats [INTELLECT-3](intellect-3.md), [MiniMax-M2](minimax-m2.md), [Nemotron 3 Super 120B-A12B](nemotron-3-super-120b-a12b.md) here, though they rank higher overall. Trails [LFM2.5 1.2B Thinking](lfm2-5-1-2b-thinking.md), [LFM2.5 8B-A1B](lfm2-5-8b-a1b.md), [LFM2.5 2.6B](lfm2-5-2-6b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
LMArena (text) 1365.3 ▁▁▁▂▃▃▃▆●▇█▃ beats 210/385 avg 1260 2.4% blunt
Arena-Hard v2.0 hard prompts (sty… 15.0% ▂·▃●▅·▃█▆▂▂▂ beats 5/22 very hard · avg 12% 1.8% sharp
AlpacaEval 2 length-controlled 63.4% ▃▃·▁▁█▇▅·●·▁ beats 21/22 medium · avg 39% 1.6% sharp · at its level
Arena-Hard v2.0 creative writing … 69.9% ▆▂▂▆▂▂▃█●▃·▂ beats 19/22 hard · avg 19% 0.5% sharp · at its level · noisy
IFEval prompt-level loose 83.4% ▁▃▁▃▁▄▁▁●▃▃▁ beats 16/22 easy · avg 78% 0.4% typical
IFEval (maker) 90.4% ●··▄··█···▄▄ beats 0/5 very easy · avg 94% 0.4% maxed out
math · #204 of 431 · 15% of its score
Beats [MiniMax-M2.1](minimax-m2-1.md), [DeepSeek V3 (Dec '24)](deepseek-v3.md), [Qwen3 235B-A22B (Apr 2025, hybrid)](qwen3-235b-a22b.md) here, though they rank higher overall. Trails [Mixtral 8x7B v0.1 (base)](mixtral-8x7b-v0-1.md), [Olmo 3 7B Instruct](olmo-3-7b-instruct.md), [Phi-4-mini-reasoning 3.8B](phi-4-mini-reasoning.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
MATH level 5 74.0% ▂▂▅▆▄█▆●▅▂▃▄ beats 63/92 hard · avg 28% 5.3% sharp · at its level
MATH (subset unspecified) 76.2% ▃▃█▂▂●·····▁ beats 20/22 easy · avg 82% 2.4% several runs
GSM8K 91.1% ▂▃··▃▂▅█▆●▂▃ beats 19/22 very easy · avg 93% 2.2% sharp · maxed out · several runs
AIME 2026 (maker) 20.8% ●·▄▄····▄··█ beats 0/5 medium · avg 40% 1.2% sharp · at its level · noisy
MATH-Vision (maker) 46.0% ●█·█·····███ beats 0/5 medium · avg 53% 1.1% at its level
OTIS mock AIME 2024–2025 22.5% ▃▃▄▃●▃▇▇█▃▂▅ beats 60/197 very hard · avg 7.2% 1.0% sharp · noisy
MathVista (testmini) (maker) 67.6% ▁··▁▄▄▆●▄▅█▄ beats 22/48 medium · avg 60% 0.9% at its level · small run
OMEGA (math generalisation) 24.0% ▅···▃●···▃▃█ beats 3/8 hard · avg 24% 0.5% at its level
vision & documents · #112 of 153 · 10% of its score
Beats [DeepSeek V4 Flash Vision (Exp)](deepseek-v4-flash-vision-exp.md), [GPT-4o (2024-08-06)](gpt-4o-2024-08-06.md), [GPT-4 Turbo (2024-04-09)](gpt-4-turbo-2024-04-09.md) here, though they rank higher overall.
test result vs the others in its run difficulty counts why
Arena vision (style controlled) 1159.9 ▁▁▂▂▂▂▃●▅▅█▇ beats 54/152 avg 1061 1.6% blunt
MMMU Pro (maker) 49.7% █●·█····█·██ beats 1/5 medium · avg 48% 1.4% at its level
VisEditBench strict pass rate (1,… 43.0% ▄··█·●··▄█·▄ beats 3/7 hard · avg 33% 1.2% at its level
InfographicVQA (maker) 70.6% ▄●·······█·▄ beats 2/5 easy · avg 70% 1.2% typical
MATH-Vision (maker) 46.0% ●█·█·····███ beats 0/5 medium · avg 53% 1.1% at its level
MMMU (val) (maker) 64.9% ▁····▁▁▃█●▆▅ beats 30/53 medium · avg 57% 0.9% at its level · small run
MathVista (testmini) (maker) 67.6% ▁··▁▄▄▆●▄▅█▄ beats 22/48 medium · avg 60% 0.9% at its level · small run
OmniDocBench 1.5 ↓ (maker) 36.5% ●·█·····████ beats 0/5 avg 31% 0.5% blunt
VMMU (Vietnamese multimodal under… 39.1% ▅······▃·●·█ beats 4/7 very hard · avg 36% 0.4% noisy
VQAv2 (val) (maker) 71.0% ▃▅▃●·▃▅·▃·▃▅ beats 5/13 easy · avg 72% 0.3% blunt · at its level · small run
ChartQA (maker) 78.0% ▁··▁▁▁▁▁▂●▅█ beats 11/44 easy · avg 77% 0.3% blunt · small run
DocVQA (maker) 86.6% ▁··●▁▁▁▇█▇▄▅ beats 1/28 very easy · avg 89% 0.2% blunt · small run
TextVQA (maker) 65.1% ▁▁●▄▅▁▃·▃▃▃█ beats 2/24 easy · avg 74% 0.1% blunt · small run
AI2D test (maker) 84.5% ▁▁▅▇▃●·▄·█·▃ beats 17/28 easy · avg 83% 0.1% blunt · small run
other tests (36)
test result vs the others in its run difficulty counts why
GPQA Diamond 47.7% ▁▁▁▁▂▄●▄▅▇█▅ beats 61/211 hard · avg 39% 6.1% sharp · at its level · several runs
MMLU 81.8% ▁▂·▁·▁▁█·▃●▅ beats 16/22 easy · avg 75% 5.0% at its level · several runs
BIG-Bench Hard 83.7% ▁▄·▁▁·▅█·▄●▁ beats 20/22 medium · avg 72% 4.8% sharp · at its level · several runs
Medmarks-V mean win rate (19 veri… 0.4610 ▁▄▃▃●▄█▃▅▁▁▁ beats 12/30 avg 0.418 4.4% sharp
Medmarks-V weighted mean win rate… 46.1% ▂▃▅▂▆●█▅█▅▃▆ beats 11/34 medium · avg 41% 2.0% blunt · at its level
MMLU-Pro (maker) 67.6% ▄··●···▄··▄▄ beats 1/5 medium · avg 60% 1.9% at its level
CPTUB average (SpeakLeash, 0-5 ju… 76.1% ▁▂▂▂▃█▃▆●▄▂▃ beats 72/95 medium · avg 65% 1.6% at its level
Polish medical board exams, bias-… 59.2% ▄·▂▂▂··█●▄·▄ beats 9/13 hard · avg 46% 1.6% at its level
MMMLU (maker) 70.7% █●··█···█·██ beats 1/5 medium · avg 69% 1.6% at its level
RobloxQA v2.0 multiple choice (To… 84.5% ▁▁●▁█▄▃▃▆▂▂▂ beats 4/49 easy · avg 77% 1.6% at its level
SimpleQA (maker) 10.0% ▄▄▄▆●▆███·▂█ beats 11/32 very hard · avg 6.8% 1.5% typical
EQ-Bench 3 (emotional intelligenc… 1159.4 ▁·▂▁▄▅●▃█▄▁▁ beats 22/62 avg 1034 1.5% typical
BRIDGE clinical text, zero-shot a… 39.9% ▁▁▁▁▁▂▂▃▄▆●▃ beats 74/97 hard · avg 31% 1.4% blunt · at its level
Global-MMLU-Lite (maker) 75.1% ▂···▂·▂●▂·██ beats 3/12 medium · avg 64% 1.4% at its level
KCL-Essay vanilla 23.6% ▁···▁●▂█▂▃▃▂ beats 4/23 hard · avg 17% 1.3% typical
HindiGen-v1 3C3H score (Inception… 50.3% ▁▁▁▁·▂·●▆█▇▁ beats 8/31 medium · avg 46% 1.3% at its level
Open Persian LLM Leaderboard v2 a… 64.6% ▂▄▂▃█▄▃▆●▅▂▂ beats 62/81 medium · avg 57% 1.1% blunt · at its level
MIMIC-CDM full information (4 pat… 75.5% ▄·······▄●·▄ beats 3/4 easy · avg 71% 1.1% typical
MIMIC-CDM clinical decision makin… 70.5% █······█··█● beats 3/3 medium · avg 60% 1.1% at its level
IFBench (maker) 32.0% ●▄·▄······▄█ beats 0/5 medium · avg 40% 1.1% at its level
MRCR v2 8-needle, 128k (maker) 13.5% ●▄·▄···█···▄ beats 0/5 hard · avg 22% 1.0% at its level
Big Bench Extra Hard micro avg (m… 19.3% ●··▄···▄·▄·▄ beats 0/5 hard · avg 28% 0.8% at its level
PopQA 30.9% ▁▁▁▃▃▃▁●▁▃·▄ beats 15/21 hard · avg 31% 0.8% at its level
RULER 128k (maker) 66.0% ▃··▅▄●▅█▇·▃▃ beats 13/33 medium · avg 59% 0.7% at its level · small run
FACTS Grounding (maker) 74.9% ▄········▄·● beats 2/3 easy · avg 72% 0.7% typical
LongSocialBench (full-context, 4-… 36.5% ●▂·▄█·▂▂·▄▄▂ beats 2/16 very hard · avg 36% 0.6% blunt
KCL-MCQA vanilla 35.1% ▂···█▂●▆██▆▂ beats 7/23 very hard · avg 29% 0.6% typical
Open PT-LLM Leaderboard average (… 78.7% ▃▁▁▁▁▁▁▁▂█●▁ beats 247/276 medium · avg 72% 0.5% at its level · noisy
EuroEval HellaSwag (English), tes… 82.1% ▃·▁▁▂▁▂▅█●▄▁ beats 77/100 medium · avg 67% 0.4% sharp · at its level · noisy
WMT24++ (chrF) (maker) 53.4% █······█··█● beats 3/3 medium · avg 52% 0.4% blunt · at its level
EuroEval Life in the UK (English … 92.1% ▁▁▁▁▁▁▁▂▄█●▃ beats 126/146 easy · avg 82% 0.2% noisy
WMT24++ (maker) 83.1% █···█·█··●██ beats 3/5 easy · avg 73% 0.2% blunt · at its level
Chess Puzzles (Epoch) 0.0% ●▁▁▁▁▂▃▃▃▂▁▁ beats 20/137 very hard · avg 5.8% 0.2% sharp · too hard for it · noisy
Vectara hallucination leaderboard… 97.0% ▁▁▂▃▄▄█●▃▂▂▁ beats 105/140 very easy · avg 94% 0.1% blunt · maxed out · noisy
TruthfulQA MC2 63.9% ▃▃▂·▃▃·▅●▂▅▂ beats 12/21 medium · avg 60% 0.1% blunt · at its level
Vectara hallucination leaderboard… 92.6% ▁▂▁▂▅█▂●▄▃▁▁ beats 67/98 very easy · avg 90% 0.0% blunt
source results share of the score
Gemma Team (submitted / maker-reported) 18 21% ▇▇
Google DeepMind Gemma 3 270M IT model c… 15 15% ▇▇
2 OLMo 2 Furious 8 10% ▇
Epoch AI 4 10% ▇
Medmarks: A Comprehensive Open-Source L… 2 6% ▇
LMArena leaderboard dataset 2 4% ▇
Gemma Team (submitted / maker-reported) 5 3% ▇
Roblox LLM Leaderboard 2 3% ▇
Aider polyglot benchmark runs 1 3% ▇
WebApp1K models leaderboard 2 2% ▇
Arena-Hard-Auto v2.0-Preview leaderboard 2 2% ▇
MIMIC Clinical Decision Making leaderbo… 2 2% ▇
Korean Canonical Legal Benchmark: Towar… 2 2% ▇
SpeakLeash CPTUB 1 2% ▇
Reassessing High-Performing LLMs on Pol… 1 2% ▇
Berkeley Function Calling Leaderboard r… 2 2% ▇
EQ-Bench 3 canonical leaderboard results 1 1% ▇
BRIDGE Medical Leaderboard 1 1% ▇
Inception Arabic/Hindi leaderboards 1 1% ▇
VisEditBench: Can Vision-Language Model… 1 1% ▇
Open Persian LLM Leaderboard v2 1 1% ▇
allenai/Olmo-3.1-32B-Think model card 2 1% ▇
google/gemma-4-31B-it model card (submi… 1 1% ▇
EuroEval leaderboards 2 1% ▇
LongSocialBench: Do Long-Context LLMs U… 1 1% ▇
Open Portuguese LLM Leaderboard 1 1% ▇
VMMU: A Vietnamese Multitask Multimodal… 1 0% ▇
TranslateGemma Technical Report (submit… 1 0% ▇
Vectara hallucination leaderboard 2 0% ▇
prior: pull toward the middle 0% ▇
weighted average = 53.4
Licences: Gemma Team ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · Google DeepMind Gemma 3 270M IT model card ([CC-BY-4.0](../../licenses/cards-final-NOTICE.txt)) · 2 OLMo 2 Furious ([CC BY 4.0](../../licenses/paper-2501.00656v3-CC-BY-4.0.txt)) · Epoch AI ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · Medmarks: A Comprehensive Open-Source LLM Benchmark Suite for Medical Tasks ([CC BY 4.0](../../licenses/paper-2605.01417v1-CC-BY-4.0.txt)) · LMArena leaderboard dataset ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · Gemma Team ([CC BY 4.0](../../licenses/report-2503.19786v1.txt)) · Roblox LLM Leaderboard ([MIT](../../licenses/roblox-llm-leaderboard-MIT.txt)) · Aider polyglot benchmark runs ([Apache-2.0](../../licenses/aider-polyglot-runs-Apache-2.0.txt)) · WebApp1K models leaderboard ([MIT](../../licenses/webapp1k-leaderboard-MIT.txt)) · Arena-Hard-Auto v2.0-Preview leaderboard ([Apache-2.0](../../licenses/arena-hard-auto-Apache-2.0.txt)) · MIMIC Clinical Decision Making leaderboard ([Apache-2.0](../../licenses/hf-leaderboard-spaces-Apache-2.0.txt)) · Korean Canonical Legal Benchmark: Toward Knowledge-Independent Evaluation of LLMs’ Legal Reasoning Capabilities ([CC BY 4.0](../../licenses/cards-final-NOTICE.txt)) · SpeakLeash CPTUB ([MIT](../../licenses/hf-leaderboard-spaces-MIT.txt)) · Reassessing High-Performing LLMs on Polish Medical Exams: True Competence or Bias-Driven Performance? ([CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/)) · Berkeley Function Calling Leaderboard result archive ([Apache-2.0](../../licenses/bfcl-result-archive-v4-Apache-2.0.txt)) · EQ-Bench 3 canonical leaderboard results ([MIT](../../licenses/eqbench3-MIT.txt)) · BRIDGE Medical Leaderboard ([MIT](../../licenses/bridge-medical-leaderboard-MIT.txt)) · Inception Arabic/Hindi leaderboards ([Apache-2.0](../../licenses/hf-leaderboard-spaces-Apache-2.0.txt)) · VisEditBench: Can Vision-Language Models Edit Visualization Code from Multimodal Feedback? ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · Open Persian LLM Leaderboard v2 ([Apache-2.0](../../licenses/hf-leaderboard-spaces-Apache-2.0.txt)) · allenai/Olmo-3.1-32B-Think model card ([Apache-2.0](../../licenses/allenai--Olmo-3.1-32B-Think.txt)) · google/gemma-4-31B-it model card ([Apache-2.0](../../licenses/google--gemma-4-31B-it.txt)) · EuroEval leaderboards ([MIT](https://github.com/EuroEval/leaderboards/blob/1a77f1d2f8ba1ba05ab73287fb090d76bf41b2e0/LICENSE)) · LongSocialBench: Do Long-Context LLMs Understand Online Discussion Threads? ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · Open Portuguese LLM Leaderboard ([Apache-2.0](../../licenses/hf-leaderboard-spaces-Apache-2.0.txt)) · VMMU: A Vietnamese Multitask Multimodal Understanding and Reasoning Benchmark ([CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/)) · TranslateGemma Technical Report ([CC BY 4.0](http://creativecommons.org/licenses/by/4.0/)) · Vectara hallucination leaderboard ([Apache-2.0](https://www.apache.org/licenses/LICENSE-2.0))
## How to run it
### Through an API
api (DeepInfra) · $0.10/M blended · $0.08 in / $0.16 out per 1M tokens · [deepinfra.com](https://deepinfra.com/google/gemma-3-27b-it)
[price source](https://github.com/BerriAI/litellm/blob/c19ce71bcdad234ca87a7af10cee45375ce811a8/model_prices_and_context_window.json): USD per 1M text tokens, standard uncached input and output. Product: Serverless text generation. API model: google/gemma-3-27b-it. Settings: service tier=standard, batch=no, cache=uncached input. Additional/excluded charges: Tools, cache storage, priority service and other modalities excluded.. Currency: USD. · [context source](https://github.com/BerriAI/litellm/blob/c19ce71bcdad234ca87a7af10cee45375ce811a8/model_prices_and_context_window.json): provider maximum input tokens; not an intrinsic model maximum or legacy max\_tokens
### 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 3 27B (Q4_K_M, needs ~16.5 GB, from [huggingface.co/ggml-org](https://huggingface.co/ggml-org/gemma-3-27b-it-GGUF/tree/f94c25afed0072339c5fa3b705a7b4222afe5f62))
```sh
llama cli -hf ggml-org/gemma-3-27b-it-GGUF:Q4_K_M
```
3. Or start a local server and open the address it prints (http://localhost:9931)
```sh
llama serve -hf ggml-org/gemma-3-27b-it-GGUF:Q4_K_M
```
Other versions: Q8_0 (~28.7 GB) `llama cli -hf ggml-org/gemma-3-27b-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 3 27B (Q4_K_M, needs ~16.5 GB, from [huggingface.co/ggml-org](https://huggingface.co/ggml-org/gemma-3-27b-it-GGUF/tree/f94c25afed0072339c5fa3b705a7b4222afe5f62))
```sh
lms get https://huggingface.co/ggml-org/gemma-3-27b-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 (~28.7 GB) `lms get https://huggingface.co/ggml-org/gemma-3-27b-it-GGUF@Q8_0 -y` · 4bit (~15.2 GB, Mac only) `lms get https://huggingface.co/Outlier-Ai/gemma-3-27b-it-MLX-4bit -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 3 27B (Q4_K_M, needs ~16.94 GB, from [ollama.com](https://ollama.com/library/gemma3/tags))
```sh
ollama run gemma3:27b
```
Other versions: qat (~16.94 GB) `ollama run gemma3:27b-it-qat` · Q8_0 (~33.38 GB) `ollama run gemma3:27b-it-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 3 27B (4bit, needs ~15.2 GB, Mac only, from [huggingface.co/Outlier-Ai](https://huggingface.co/Outlier-Ai/gemma-3-27b-it-MLX-4bit/tree/72a610927c5abd65bf1e6159caa4e28e69fca85e))
```sh
mlx_lm.chat --model Outlier-Ai/gemma-3-27b-it-MLX-4bit
```
3. Or start an OpenAI-compatible server at http://localhost:8080/v1
```sh
mlx_lm.server --model Outlier-Ai/gemma-3-27b-it-MLX-4bit
```
Model repository and documentation: [huggingface.co/google](https://huggingface.co/google/gemma-3-27b-it)
[interactive version](https://whatmodel.sh/model/gemma-3-27b) · [JSON](https://whatmodel.sh/api?format=json&all=1)