Google
ChatCodingReasoningMulti-purpose27.23B
Chat

Gemma 3 27B

Gemma · Gemma Terms of Use

Google's largest open-weight model before Gemma 4. Capable reasoning and instruction following. At 27B parameters, it sits between 8B models (too limited) and 70B models (too expensive). Wide multilingual support. Fits on 24 GB GPUs at Q4.

Parameters
27.23B
Architecture
Dense
Context
128,000 tokens
Released
2025-03-12
Engines
llama.cpp, ollama, vLLM
Builder Tools
Cursor, Continue, LM Studio, Open WebUI

Parameters

27.23B

VRAM

22.3 GB

Context

125K

Formats

5

GPUs

42

Gemma 3 27B (27.23B) requires 22.3 GB VRAM at recommended quality (Q6_K). At efficient quality (Q4_K_M), it fits in 16.3 GB VRAM, making it compatible with the NVIDIA RTX 4090 Laptop (150-175W). On NVIDIA Grace Blackwell Ultra GB300, expect approximately 130 tok/s at Q8_0. For the best experience, AMD AI Powerhouse ($1,818) is recommended.

Source: OwnRig methodology

VRAM (Recommended)

22.3 GB

Quantization

Q6_K

File Size

20.4 GB

Max Context

125K tokens

Primary Use

Chat

Memory

VRAM Requirements

QualityQuantizationVRAMFile Size
fullQ8_029.5 GB27.2 GB
recommendedQ6_K22.3 GB20.4 GB
recommendedQ5_K_M19.3 GB17.5 GB
efficientQ4_K_M16.3 GB14.8 GB
compressedQ3_K_M13.3 GB12.1 GB
Scaling

Context Length Impact

KV cache VRAM at Q6_K quality. Longer context = more memory.

ContextKV CacheTotal VRAM
2K307 MB22.6 GB
4K512 MB22.8 GB
8K1 GB23.3 GB
16K2 GB24.3 GBexceeds 24 GB
32K4.1 GB26.4 GBexceeds 24 GB
64K8.2 GB30.5 GBexceeds 24 GB

Compatible GPUs

42 devices
NVIDIA Grace Blackwell Ultra GB300Q8_0130 tok/sExcellent
NVIDIA GeForce RTX 5090Q5_K_M35 tok/sExcellent
Apple M4 Max (128GB Unified)Q8_012 tok/sGood
Apple M4 Max (36GB Unified)Q5_K_M15 tok/sGood
Apple M4 Max (64GB Unified)Q6_K14 tok/sGood
Apple M4 Ultra (192GB)Q8_018 tok/sGood
NVIDIA GeForce RTX 3090Q4_K_M18 tok/sGood
NVIDIA GeForce RTX 4090Q4_K_M22 tok/sGood
AMD Radeon RX 7900 XTXQ4_K_M15 tok/sGood
NVIDIA RTX PRO 6000 BlackwellQ8_036 tok/sGood
NVIDIA RTX PRO 6000 Blackwell Max-QQ8_033 tok/sGood
Apple M4 Pro (24GB Unified)Q4_K_M8 tok/sAcceptable
Apple M4 Pro (48GB)Q5_K_M8 tok/sAcceptable
NVIDIA GeForce RTX 4070 Ti SuperQ3_K_M12 tok/sAcceptable
NVIDIA GeForce RTX 4080 SuperQ3_K_M14 tok/sAcceptable
NVIDIA GeForce RTX 5080Q3_K_M18 tok/sAcceptable
AMD Radeon Pro W7900Q5_K_M9 tok/sAcceptable
AMD Radeon RX 9070Q3_K_M10 tok/sAcceptable
Apple M3 Pro (18GB Unified)Q3_K_M3 tok/sMarginal
NVIDIA GeForce RTX 4060 Ti 16GBQ3_K_M6 tok/sMarginal
NVIDIA RTX 4090 Laptop (150-175W)Q3_K_M5 tok/sMarginal
AMD Radeon RX 7600Q3_K_M2 tok/sMarginal
AMD Radeon RX 9060 XT 16GBQ3_K_M5 tok/sMarginal
NVIDIA GeForce RTX 5060 Ti 16GBQ3_K_M7 tok/sMarginal
Apple M4 (16GB Unified)Q4_K_MNot viable
NVIDIA GeForce RTX 3060 12GBQ3_K_MNot viable
NVIDIA GeForce RTX 3080 10GBQ3_K_MNot viable
NVIDIA GeForce RTX 4060 8GBQ3_K_MNot viable
NVIDIA RTX 4060 Laptop (40-60W)Q3_K_MNot viable
NVIDIA RTX 4070 Laptop (80-115W)Q3_K_MNot viable
NVIDIA GeForce RTX 4070 SuperQ3_K_MNot viable
NVIDIA GeForce RTX 4070 Ti 12GBQ3_K_MNot viable
Apple M1 (8GB Unified)Q4_K_MNot viable
Apple M1 (16GB Unified)Q4_K_MNot viable
Apple M1 Pro (16GB Unified)Q4_K_MNot viable
Apple M2 (8GB Unified)Q4_K_MNot viable
Apple M2 (16GB Unified)Q4_K_MNot viable
Apple M2 Pro (16GB Unified)Q4_K_MNot viable
Apple M3 (8GB Unified)Q4_K_MNot viable
Apple M3 (16GB Unified)Q4_K_MNot viable
AMD Radeon RX 9060 XT 8GBQ3_K_MNot viable
NVIDIA GeForce RTX 5060 8GBQ3_K_MNot viable

Showing 42 of 42 entries

Builder Context

Gemma 3 27B is commonly used with Cursor, Continue, LM Studio, Open WebUI. For an AI coding workflow, pair it with an embedding model like nomic-embed-text for local RAG.

Hardware

Recommended Builds

Complete PC builds that can run Gemma 3 27B.

FAQ

Frequently Asked Questions

How much VRAM does Gemma 3 27B need?
Gemma 3 27B requires 22.3 GB VRAM at recommended quality (Q6_K). At lower quality settings, it can fit in as little as 13.3 GB.
What is the best GPU for Gemma 3 27B?
The NVIDIA Grace Blackwell Ultra GB300 delivers the best performance for Gemma 3 27B, achieving 130 tok/s at Q8_0 with an excellent rating.
Can I run Gemma 3 27B on an RTX 4060 Ti?
Yes. On the NVIDIA GeForce RTX 4060 Ti 16GB, Gemma 3 27B runs at 6 tok/s (Q3_K_M, marginal).
What quantization should I use for Gemma 3 27B?
For the best quality, use Q6_K (22.3 GB VRAM). If your GPU has limited VRAM, Q3_K_M (13.3 GB) is the most efficient option with acceptable quality.
Is Gemma 3 27B good for coding?
Yes. Gemma 3 27B is used with Cursor, Continue, LM Studio, Open WebUI for local AI coding. For the best coding experience, pair it with an embedding model for local RAG.

Related Guides

All models

Data confidence: estimated. Source

VRAM requirements are calculated from model parameters and may vary by inference engine, context length, and batch size. Performance estimates are based on community benchmarks and should be verified for your specific configuration.Gemma is a trademark of its respective owner. OwnRig is not affiliated with or endorsed by the model creator.