16 GB CPU desktop
Comfortable with 3B–8B Q4 models; 12B is possible only with reduced context and patience.
Qwen3 14B
The highest listed result may be a tight or offloaded fit. For daily use, prefer the first result marked Confortable and keep context modest.
Ollama
One-command local chat and app integration. GPU and driver support varies by operating system. Confirm the current hardware matrix before buying hardware. Since v0.34.2 (2026-09-15) Ollama's first-run setup offered account sign-in, and the release notes said you could continue locally; since v0.40.1 (2026-10-07) the CLI Welcome screen completes onboarding on Enter and the intermediate account step is gone, while cloud models keep the sign-in flow. Since v0.40.0 (2026-09-25) Ollama runs MLX-supported model architectures on Apple silicon through MLX by default rather than its own runner; NVIDIA and CPU paths are unchanged.
Ouvrir le guideModèles open-weight représentatifs
Qwen3 14B
- Mémoire estimée
- ~11.2 GB
- Contexte
- 32K+
Gemma 3 12B
- Mémoire estimée
- ~9.4 GB
- Contexte
- 128K
Qwen3 8B
- Mémoire estimée
- ~6.8 GB
- Contexte
- 32K+
Qwen3 4B
- Mémoire estimée
- ~3.6 GB
- Contexte
- 32K+
Gemma 3 4B
- Mémoire estimée
- ~4.2 GB
- Contexte
- 128K
Llama 3.2 3B
- Mémoire estimée
- ~2.8 GB
- Contexte
- 128K
Fit is based on estimated total model memory. Driver support, KV cache, multimodal projectors, concurrency, and desktop applications can all reduce available headroom. Current fit: Qwen3 14B (Serré), Gemma 3 12B (Serré), Qwen3 8B (Confortable), Qwen3 4B (Confortable), Gemma 3 4B (Confortable), Llama 3.2 3B (Confortable).