Can MacBook Air M5 16 GB run Qwen3 8B?
16 GB unified · 153 GB/s memory bandwidth · Qwen3 8B has 8.2B parameters.
Short answer
Yes, comfortable for chat; doesn't fit for coding in a repository.
Best quantization for chat: 4-bit (4.4 GB of weights) on LM Studio. Generation: 15–25 tok/s, leaving about 2.0 GB free for your other apps.
How it rates for each workload
Same model and machine, different jobs. Agents and long documents send far bigger prompts than chat, so they need much more speed and memory.
| Workload | Verdict | Quant | Generation | First reply | Context | |
|---|---|---|---|---|---|---|
| Chatwith Open WebUI | Comfortable | 4-bit | 15–25 tok/s | ≈1 s | 8K | Details |
| Coding questionswith Continue | Borderline | 3-bit | 17–30 tok/s | ≈9 s | 16K | Details |
| Coding in a repositorywith Aider | Does Not Fit | — | — | — | — | Details |
| Agentic codingwith OpenCode | Does Not Fit | — | — | — | — | Details |
| Long documents (64K)with Open WebUI | Borderline | 3-bit | 9–16 tok/s | ≈75 s | 32K | Details |
Speeds are bandwidth-based estimates; no direct benchmark exists for this pair. “First reply” is the time to read the workload’s opening prompt. How we calculate
Get it running
Copy-paste commands for the 4-bit build with a 8K context. Ollama and llama.cpp download the file straight from Hugging Face.
ollama run hf.co/unsloth/Qwen3-8B-GGUF:Q4_K_M
Downloads 5.03 GB (Q4_K_M). Ollama's default context is small: set OLLAMA_CONTEXT_LENGTH=8192 before `ollama serve` for long prompts.
Other models for MacBook Air M5 16 GB
The best-rated models for chat on this machine.
- MiniCPM5 2B25–45 tok/sExcellent
- Qwen3.5 4B19–30 tok/sExcellent
- Llama 3.2 3B Instruct25–45 tok/sExcellent
- Gemma 4 E4B16–25 tok/sExcellent
- Llama 3.1 8B Instruct15–25 tok/sExcellent
- Qwen2.5-Coder 7B Instruct17–30 tok/sExcellent
Cheapest machines for Qwen3 8B
Hardware rated comfortable or better for chat, lowest price first.
Different tool, context size or workload?
These pages use typical settings. The full check models your exact setup.