Alibaba (Qwen) · Apache 2.0
Qwen3.6
April 2026 Qwen focused on coding agents: a fast 35B-A3B and a 27B dense, both reading images.
How much memory it needs
The model file, plus memory for the conversation, plus about 1 GB for the app. Longer conversations need more.
| Size | Version | File | At 8K | At 32K | At 128K |
|---|---|---|---|---|---|
| 28B | Q4_K_M | 16.8 GB | 19.3 GB | 20.9 GB | 27.4 GB |
| Q8_0 | 28.6 GB | 31.1 GB | 32.7 GB | 39.1 GB | |
| 36B (3B active) | UD-IQ4_XS | 17.7 GB | 19.8 GB | 20.3 GB | 22.3 GB |
| Q4_K_M | 22.3 GB | 24.4 GB | 24.9 GB | 26.9 GB | |
| Q8_0 | 37.8 GB | 39.9 GB | 40.4 GB | 42.4 GB |
Which computers run it
The best size for each machine at an 8K-token conversation. Speeds are estimates from memory bandwidth.
| Computer | Best version | Fit | Speed |
|---|---|---|---|
| Laptop or PC, no graphics card, 32 GB | 36B UD-IQ4_XS | Runs well | 20–40tok/sSmooth |
| RTX 4060 · 8 GB | 28B Q4_K_M | Spills into system RAM | 4–6tok/sSlow |
| RTX 3060 · 12 GB | 28B Q4_K_M | Spills into system RAM | 5.5–8.5tok/sReadable |
| RTX 5060 Ti · 16 GB | 28B Q4_K_M | Spills into system RAM | 8.5–14tok/sReadable |
| RTX 4090 · 24 GB | 36B UD-IQ4_XS | Runs well | 120–240tok/sFast |
| RTX 5090 · 32 GB | 36B Q4_K_M | Runs well | 150–290tok/sFast |
| Mac with M4 · 16 GB | 28B Q4_K_M | Too big | — |
| Mac with M4 Pro · 48 GB | 36B Q4_K_M | Runs great | 40–90tok/sFast |
| Mac with M5 Max · 128 GB | 36B Q8_0 | Runs great | 50–110tok/sFast |
| Ryzen AI Max+ 395 · 128 GB | 36B Q8_0 | Runs great | 25–55tok/sFast |
| DGX Spark · 128 GB | 36B Q8_0 | Runs great | 30–55tok/sFast |
| RTX PRO 6000 · 96 GB | 36B Q8_0 | Runs great | 120–230tok/sFast |
| Mac Studio with M5 Ultra · 256 GB | 36B Q8_0 | Runs great | 50–110tok/sFast |
| Mac Studio with M3 Ultra · 512 GB | 36B Q8_0 | Runs great | 40–85tok/sFast |
What it is good and bad at
Good at
- 35B-A3B is fast: only 3B parameters work per word
- Steady in long coding-agent sessions
Not good at
- 35B-A3B needs about 23 GB at Q4
- The 27B is replaced by Qwen3.8 27B; the 35B-A3B has no Qwen3.8 successor yet
Start it
Install Ollama, then run one of these in a terminal. LM Studio users can search the same name in the app.
ollama run qwen3.6:27bollama run qwen3.6:35b-a3b