Alibaba Cloud
Qwen 3 — all variants, VRAM & hardware
The original Qwen 3 family from Alibaba — dense 0.6B through 32B + MoE 30B-A3B and 235B-A22B. Succeeded by Qwen 3.5 in early 2026 but still widely deployed.
Variant comparison
| Variant | Params | Active | VRAM @ Q4 | Architecture | |
|---|---|---|---|---|---|
| Qwen 3 0.6B | 0.6000000238418579B | 0.6000000238418579B | ~0.4 GB | dense | Details → |
| Qwen 3 1.7B | 1.7000000476837158B | 1.7000000476837158B | ~1.0 GB | dense | Details → |
| Qwen 3 4B | 4B | 4B | ~2.4 GB | dense | Details → |
| Qwen 3 8B | 8B | 8B | ~4.9 GB | dense | Details → |
| Qwen 3 14B | 14B | 14B | ~8.5 GB | dense | Details → |
| Qwen3-VL 30B A3B Instruct | 30B | 3B | ~18.3 GB | moe | Details → |
| Qwen 3 30B A3B | 30.5B | 3.299999952316284B | ~18.6 GB | moe | Details → |
| Qwen3-Coder 30B A3B Instruct | 30.5B | 3.299999952316284B | ~18.6 GB | moe | Details → |
| Qwen 3 32B | 32B | 32B | ~19.5 GB | dense | Details → |
| Qwen3-Coder-Next | 80B | 3B | ~48.8 GB | moe | Details → |
| Qwen 3 235B A22B | 235B | 22B | ~143.4 GB | moe | Details → |
| Qwen3-Coder 480B A35B Instruct | 480B | 35B | ~292.8 GB | moe | Details → |
Deep-dive guides
Qwen 3 VRAM & hardware guides
FAQ
Frequently asked questions
What is Qwen 3?
Qwen 3 is Alibaba Cloud's the original qwen 3 family from alibaba — dense 0.6b through 32b + moe 30b-a3b and 235b-a22b. succeeded by qwen 3.5 in early 2026 but still widely deployed.
How many variants are in the Qwen 3 family?
We track 12 canonical Qwen 3 variants in our catalog, spanning 0.6000000238418579B to 480B parameters.
Which Qwen 3 variant should I run?
Pick the largest variant that fits on your hardware at Q4_K_M quantization. For 8 GB GPUs, the smallest dense variants. For 24 GB GPUs, the 27B dense or 35B MoE (if available). For workstation / Mac Studio, the 122B or larger MoE. Use our fit calculator on each model page for the exact verdict.
See also