Xiaomi · MIT
MiMo-V2.6 9B
Xiaomi’s small local edition of MiMo-V2.6: Qwen3.5 9B trained further on agent and coding data.
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 |
|---|---|---|---|---|---|
| 9.4B | Q4_K_M | 5.8 GB | 8.0 GB | 8.8 GB | 12.0 GB |
| Q8_0 | 9.6 GB | 11.7 GB | 12.5 GB | 15.8 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 | 9.4B Q8_0 | Runs great | 5.5–8.5tok/sReadable |
| RTX 4060 · 8 GB | 9.4B Q4_K_M | Spills into system RAM | 25–40tok/sSmooth |
| RTX 3060 · 12 GB | 9.4B Q4_K_M | Runs great | 35–55tok/sFast |
| RTX 5060 Ti · 16 GB | 9.4B Q8_0 | Runs great | 25–45tok/sSmooth |
| RTX 4090 · 24 GB | 9.4B Q8_0 | Runs great | 60–90tok/sFast |
| RTX 5090 · 32 GB | 9.4B Q8_0 | Runs great | 100–160tok/sFast |
| Mac with M4 · 16 GB | 9.4B Q4_K_M | Runs great | 11–19tok/sSmooth |
| Mac with M4 Pro · 48 GB | 9.4B Q8_0 | Runs great | 16–25tok/sSmooth |
| Mac with M5 Max · 128 GB | 9.4B Q8_0 | Runs great | 35–60tok/sFast |
| Ryzen AI Max+ 395 · 128 GB | 9.4B Q8_0 | Runs great | 16–25tok/sSmooth |
| DGX Spark · 128 GB | 9.4B Q8_0 | Runs great | 17–25tok/sSmooth |
| RTX PRO 6000 · 96 GB | 9.4B Q8_0 | Runs great | 100–160tok/sFast |
| Mac Studio with M5 Ultra · 256 GB | 9.4B Q8_0 | Runs great | 50–90tok/sFast |
| Mac Studio with M3 Ultra · 512 GB | 9.4B Q8_0 | Runs great | 35–60tok/sFast |
What it is good and bad at
Good at
- Coding and agent tasks at 9B
Not good at
- A research starting point, not a polished assistant
- No Ollama library entry