Qwen Models: Local AI on Any Device

Qwen by Alibaba Cloud offers the most versatile range of open-weight models available today. With variants from 0.5B to 235B parameters, there is a Qwen model for every device, from iPhones with 6GB RAM to Mac Studios with 192GB. Qwen consistently scores among the top models for coding, chat, and multilingual tasks.

Alibaba Cloud28 local models
DEVELOPER
Alibaba Cloud
MODELS
28
SIZE RANGE
0.5B–235B
RAM RANGE
2–192 GB
Bar chart: maximum local LLM size by memory tier. 8 GB runs up to 9B, 12 GB runs up to 12B, 16 GB runs up to 14B, 24 GB runs up to 27B, 32 GB runs up to 35B, 36 GB runs up to 35B, 48 GB runs up to 35B, 64 GB runs up to 70B, 72 GB runs up to 70B, 96 GB runs up to 70B, 128 GB runs up to 70B, 192 GB runs up to 70B, 256 GB runs up to 70B, 512 GB runs up to 405B. Data from ModelFit's own catalog.
Key Features
Widest size range (0.5B to 235B)
Strong multilingual and coding performance
Best quality/size ratio in the 7B-14B range
Works on devices from iPhones to Mac Studios

All Qwen Models

ModelSizeQuantVRAMMin RAMBest ForQualityOllama
Qwen2.5 0.5B Instruct0.5BQ4_K_M0.8 GB2 GBChat, Mobile
42
Qwen3.5 0.8B Instruct0.8BQ4_K_M0.8 GB2 GBChat, Mobile
56
Qwen2.5 1.5B Instruct1.5BQ4_K_M1.5 GB4 GBChat, Translation
55
Qwen3.5 2B Instruct2BQ4_K_M1.8 GB4 GBChat, Edge tasks
70
Qwen2.5 3B Instruct3BQ4_K_M2.5 GB4 GBChat, Coding
62
Qwen3.5 4B Instruct4BQ4_K_M3.5 GB6 GBCoding, Agents, Multimodal
83
Qwen2.5 7B Instruct7BQ4_K_M5.5 GB8 GBChat, Coding
74
Qwen2.5 Coder 7B7BQ4_K_M5.5 GB8 GBCoding
78
Qwen3 8B8BQ4_K_M6.5 GB12 GBChat, Coding
84
Qwen3.5 9B Instruct9BQ4_K_M7 GB12 GBQuality, Coding, Reasoning
89
Qwen3.5 9B Instruct (Q8)9BQ8_010.7 GB16 GBQuality, Coding, Reasoning
91
Qwen2.5 14B Instruct14BQ4_K_M11 GB16 GBCoding, Chat
82
Qwen2.5 Coder 14B14BQ4_K_M11 GB16 GBCoding
85
Qwen3 14B14BQ4_K_M11 GB16 GBCoding, Quality
88
Qwen3 14B (Q8)14BQ8_015.9 GB24 GBCoding, Quality
90
Qwen3.5 27B Instruct27BQ4_K_M16 GB24 GBChat, Coding, Complex reasoning
91
Qwen3.6 27B27BQ4_K_M18 GB32 GBCoding, Quality, Long context
94
Qwen3.8 27B27BQ4_K_M16.5 GB24 GBCoding, Agent, Vision, Long context
94
Qwen3.6 27B (Q8)27BQ8_030 GB48 GBCoding, Quality, Long context
96
Qwen3 30B30BQ4_K_M22 GB32 GBQuality, Coding
89
Qwen3.5 35B-A3B Instruct35BQ4_K_M20 GB32 GBReasoning, Coding, Agent scenarios
93
Qwen3.6 35B-A3B35BQ4_K_M22 GB32 GBReasoning, Coding, Agents
95
Qwen3.6 35B-A3B (Q8)35BQ8_038.7 GB64 GBReasoning, Coding, Agents
97
Qwen3.5 35B-A3B Instruct (Q8)35BQ8_038.7 GB64 GBReasoning, Coding, Agent scenarios
95
Qwen3-Next 80B-A3B80BQ4_K_M50.4 GB72 GBChat, Coding, Long Context
94
Qwen3-Next 80B-A3B (Q8)80BQ8_084.8 GB128 GBChat, Coding, Long Context
96
Qwen3.5 122B-A10B Instruct122BQ4_K_M72 GB96 GBFrontier-level reasoning, Complex tasks
97
Qwen3 235B A22B235BQ4_K_M130 GB192 GBQuality, Reasoning
98

Device Compatibility

Which Qwen models can run on each device class, based on minimum RAM requirements.

ModeliPhoneAirProStudioMini
Qwen2.5 0.5B Instruct (0.5B)ExcellentExcellentExcellentExcellentExcellent
Qwen3.5 0.8B Instruct (0.8B)ExcellentExcellentExcellentExcellentExcellent
Qwen2.5 1.5B Instruct (1.5B)ExcellentExcellentExcellentExcellentExcellent
Qwen3.5 2B Instruct (2B)ExcellentExcellentExcellentExcellentExcellent
Qwen2.5 3B Instruct (3B)ExcellentExcellentExcellentExcellentExcellent
Qwen3.5 4B Instruct (4B)PossiblePossibleExcellentExcellentExcellent
Qwen2.5 7B Instruct (7B)PossiblePossibleExcellentExcellentExcellent
Qwen2.5 Coder 7B (7B)PossiblePossibleExcellentExcellentExcellent
Qwen3 8B (8B)PossiblePossiblePossibleExcellentPossible
Qwen3.5 9B Instruct (9B)PossiblePossiblePossibleExcellentPossible
Qwen3.5 9B Instruct (Q8) (9B)NoPossiblePossibleExcellentPossible
Qwen2.5 14B Instruct (14B)NoPossiblePossibleExcellentPossible
Qwen2.5 Coder 14B (14B)NoPossiblePossibleExcellentPossible
Qwen3 14B (14B)NoPossiblePossibleExcellentPossible
Qwen3 14B (Q8) (14B)NoPossiblePossiblePossiblePossible
Qwen3.5 27B Instruct (27B)NoPossiblePossiblePossiblePossible
Qwen3.6 27B (27B)NoPossiblePossiblePossiblePossible
Qwen3.8 27B (27B)NoPossiblePossiblePossiblePossible
Qwen3.6 27B (Q8) (27B)NoNoPossiblePossiblePossible
Qwen3 30B (30B)NoPossiblePossiblePossiblePossible
Qwen3.5 35B-A3B Instruct (35B)NoPossiblePossiblePossiblePossible
Qwen3.6 35B-A3B (35B)NoPossiblePossiblePossiblePossible
Qwen3.6 35B-A3B (Q8) (35B)NoNoPossiblePossiblePossible
Qwen3.5 35B-A3B Instruct (Q8) (35B)NoNoPossiblePossiblePossible
Qwen3-Next 80B-A3B (80B)NoNoPossiblePossibleNo
Qwen3-Next 80B-A3B (Q8) (80B)NoNoPossiblePossibleNo
Qwen3.5 122B-A10B Instruct (122B)NoNoPossiblePossibleNo
Qwen3 235B A22B (235B)NoNoNoPossibleNo

RAM Requirements

Qwen2.5 0.5B Instruct
0.8 GB · min 2 GB
Qwen3.5 0.8B Instruct
0.8 GB · min 2 GB
Qwen2.5 1.5B Instruct
1.5 GB · min 4 GB
Qwen3.5 2B Instruct
1.8 GB · min 4 GB
Qwen2.5 3B Instruct
2.5 GB · min 4 GB
Qwen3.5 4B Instruct
3.5 GB · min 6 GB
Qwen2.5 7B Instruct
5.5 GB · min 8 GB
Qwen2.5 Coder 7B
5.5 GB · min 8 GB
Qwen3 8B
6.5 GB · min 12 GB
Qwen3.5 9B Instruct
7 GB · min 12 GB
Qwen3.5 9B Instruct (Q8)
10.7 GB · min 16 GB
Qwen2.5 14B Instruct
11 GB · min 16 GB
Qwen2.5 Coder 14B
11 GB · min 16 GB
Qwen3 14B
11 GB · min 16 GB
Qwen3 14B (Q8)
15.9 GB · min 24 GB
Qwen3.5 27B Instruct
16 GB · min 24 GB
Qwen3.6 27B
18 GB · min 32 GB
Qwen3.8 27B
16.5 GB · min 24 GB
Qwen3.6 27B (Q8)
30 GB · min 48 GB
Qwen3 30B
22 GB · min 32 GB
Qwen3.5 35B-A3B Instruct
20 GB · min 32 GB
Qwen3.6 35B-A3B
22 GB · min 32 GB
Qwen3.6 35B-A3B (Q8)
38.7 GB · min 64 GB
Qwen3.5 35B-A3B Instruct (Q8)
38.7 GB · min 64 GB
Qwen3-Next 80B-A3B
50.4 GB · min 72 GB
Qwen3-Next 80B-A3B (Q8)
84.8 GB · min 128 GB
Qwen3.5 122B-A10B Instruct
72 GB · min 96 GB
Qwen3 235B A22B
130 GB · min 192 GB

Frequently Asked Questions

What is the best Qwen model for 16GB RAM?
Qwen2.5 7B Q4 is the sweet spot for 16GB RAM. It uses about 5.5GB, leaving room for your OS and apps. For coding tasks, it performs close to models twice its size.
Can I run Qwen on an iPhone?
Yes. Qwen2.5 0.5B and 1.5B models run on iPhones with 6GB+ RAM. They are fast enough for basic chat and translation, though quality is limited by their small size.
How does Qwen compare to Llama 3?
Qwen offers more size options (0.5B to 235B vs Llama's 1B to 405B). At 7B, Qwen2.5 edges out Llama 3.2 on coding benchmarks, while Llama tends to be stronger at general reasoning.
What Ollama command runs Qwen?
Run `ollama run qwen2.5:7b-instruct-q4_K_M` for the most popular size. Replace `7b` with `1.5b`, `14b`, `32b`, or `72b` depending on your RAM.

Related Model Families

LlamaMetaDeepSeekDeepSeek AIMistralMistral AI

Getting Started

Best LLM for MacBookHow to Set Up OllamaBrowse All DevicesModelFit Wizard