Qwen
ChatCodingAI codingReasoningMulti-purpose27B
Chat

Qwen3.5-27B

Qwen · Apache 2.0

Dense 27B from the Qwen 3.5 family. Native multimodal (text, image, video). 256K context window. Competitive with Gemma 3 27B and Qwen 3 32B at lower VRAM. Apache 2.0.

Parameters
27B
Architecture
Dense
Context
262,144 tokens
Released
2026-02-24
Engines
llama.cpp, ollama, vLLM
Builder Tools
Continue, LM Studio, Open WebUI

Parameters

27B

VRAM

19 GB

Context

256K

Formats

4

GPUs

43

Qwen3.5-27B (27B) requires 19 GB VRAM at recommended quality (Q5_K_M). At efficient quality (Q4_K_M), it fits in 16 GB VRAM, making it compatible with the NVIDIA RTX 4090 Laptop (150-175W). On NVIDIA Grace Blackwell Ultra GB300, expect approximately 150 tok/s at Q8_0. For the best experience, AMD AI Powerhouse ($1,818) is recommended.

Source: OwnRig methodology

VRAM (Recommended)

19 GB

Quantization

Q5_K_M

File Size

18 GB

Max Context

256K tokens

Primary Use

Chat

Memory

VRAM Requirements

QualityQuantizationVRAMFile Size
fullQ8_029 GB27 GB
recommendedQ5_K_M19 GB18 GB
efficientQ4_K_M16 GB15 GB
compressedQ3_K_M14 GB13 GB
Scaling

Context Length Impact

KV cache VRAM at Q5_K_M quality. Longer context = more memory.

ContextKV CacheTotal VRAM
2K205 MB19.2 GB
4K410 MB19.4 GB
8K819 MB19.8 GB
16K1.6 GB20.6 GB
32K3.2 GB22.2 GB
64K6.4 GB25.4 GBexceeds 24 GB
128K12.8 GB31.8 GBexceeds 24 GB

Compatible GPUs

43 devices
NVIDIA Grace Blackwell Ultra GB300Q8_0150 tok/sExcellent
Apple M4 Max (128GB Unified)Q8_016 tok/sExcellent
Apple M4 Max (64GB Unified)Q8_016 tok/sExcellent
Apple M4 Ultra (192GB)Q8_024 tok/sExcellent
NVIDIA GeForce RTX 3090Q5_K_M24 tok/sGood
NVIDIA GeForce RTX 4090Q5_K_M40 tok/sGood
NVIDIA GeForce RTX 5090Q8_039 tok/sGood
AMD Radeon RX 7900 XTXQ5_K_M21 tok/sGood
AMD Radeon Pro W7900Q8_017 tok/sGood
NVIDIA RTX PRO 6000 BlackwellQ8_036 tok/sGood
NVIDIA RTX PRO 6000 Blackwell Max-QQ8_033 tok/sGood
AMD Radeon RX 9070Q3_K_M44 tok/sGood
Apple M1 Pro (16GB Unified)Q3_K_M18 tok/sGood
Apple M2 Pro (16GB Unified)Q3_K_M20 tok/sGood
Apple M3 Pro (18GB Unified)Q4_K_M16 tok/sAcceptable
Apple M4 (16GB Unified)Q3_K_M20 tok/sAcceptable
Apple M4 Max (36GB Unified)Q8_016 tok/sAcceptable
Apple M4 Pro (24GB Unified)Q5_K_M18 tok/sAcceptable
Apple M4 Pro (48GB)Q8_09 tok/sAcceptable
NVIDIA GeForce RTX 4060 Ti 16GBQ3_K_M25 tok/sAcceptable
NVIDIA GeForce RTX 4070 Ti SuperQ3_K_M32 tok/sAcceptable
NVIDIA GeForce RTX 4080 SuperQ3_K_M38 tok/sAcceptable
NVIDIA RTX 4090 Laptop (150-175W)Q3_K_M30 tok/sAcceptable
NVIDIA GeForce RTX 5080Q3_K_M45 tok/sAcceptable
Apple M2 (16GB Unified)Q3_K_M10 tok/sAcceptable
Apple M3 (16GB Unified)Q3_K_M10 tok/sAcceptable
AMD Radeon RX 9060 XT 16GBQ3_K_M22 tok/sAcceptable
NVIDIA GeForce RTX 5060 Ti 16GBQ3_K_M28 tok/sAcceptable
NVIDIA GeForce RTX 3060 12GBQ3_K_M5 tok/sMarginal
NVIDIA GeForce RTX 4070 SuperQ3_K_M9 tok/sMarginal
NVIDIA GeForce RTX 4070 Ti 12GBQ3_K_M8 tok/sMarginal
NVIDIA RTX 4080 Laptop (120-150W)Q3_K_M8 tok/sMarginal
AMD Radeon RX 7600Q3_K_M2 tok/sMarginal
Apple M1 (16GB Unified)Q3_K_M5 tok/sMarginal
NVIDIA GeForce RTX 3080 10GBQ3_K_M7 tok/sNot 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
Apple M1 (8GB Unified)Q3_K_MNot viable
Apple M2 (8GB Unified)Q3_K_MNot viable
Apple M3 (8GB Unified)Q3_K_MNot viable
AMD Radeon RX 9060 XT 8GBQ3_K_MNot viable
NVIDIA GeForce RTX 5060 8GBQ3_K_MNot viable

Showing 43 of 43 entries

Builder Context

Qwen3.5-27B is commonly used with Continue, LM Studio, Open WebUI. For an AI coding workflow, pair it with an embedding model like nomic-embed-text for local RAG.

FAQ

Frequently Asked Questions

How much VRAM does Qwen3.5-27B need?
Qwen3.5-27B requires 19 GB VRAM at recommended quality (Q5_K_M). At lower quality settings, it can fit in as little as 14 GB.
What is the best GPU for Qwen3.5-27B?
The NVIDIA Grace Blackwell Ultra GB300 delivers the best performance for Qwen3.5-27B, achieving 150 tok/s at Q8_0 with an excellent rating.
Can I run Qwen3.5-27B on an RTX 4060 Ti?
Yes. On the NVIDIA GeForce RTX 4060 Ti 16GB, Qwen3.5-27B runs at 25 tok/s (Q3_K_M, acceptable).
What quantization should I use for Qwen3.5-27B?
For the best quality, use Q5_K_M (19 GB VRAM). If your GPU has limited VRAM, Q3_K_M (14 GB) is the most efficient option with acceptable quality.
Is Qwen3.5-27B good for coding?
Yes. Qwen3.5-27B is used with Continue, LM Studio, Open WebUI for local AI coding. For the best coding experience, pair it with an embedding model for local RAG.
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.Qwen is a trademark of its respective owner. OwnRig is not affiliated with or endorsed by the model creator.