Local LLM model fit
Mixtral 8x7B is a 46.7B Mixtral model. This page estimates Q4 VRAM fit, Ollama command, context planning, and fallback choices for common local AI GPUs.
MoE local reasoning/chat when enough VRAM is available. Weakness: Not practical for 24GB single-GPU setups without offload.
| Hardware | Examples | Clean capacity | Q4 need | Status | Calculator |
|---|---|---|---|---|---|
| 6 GB VRAM entry GPU | GTX 1660, RTX 2060 6GB | 4.5 GB usable VRAM | 28 GB | Too large | Open calculator |
| 8 GB VRAM mainstream GPU | RTX 3060 Ti, RTX 4060, RTX 3070 | 6.5 GB usable VRAM | 28 GB | Too large | Open calculator |
| 10 GB VRAM older high-end GPU | RTX 3080 10GB | 8.5 GB usable VRAM | 28 GB | Too large | Open calculator |
| 12 GB VRAM local agent GPU | RTX 3060 12GB, RTX 4070, RTX 5070 | 10.5 GB usable VRAM | 28 GB | RAM offload | Open calculator |
| 16 GB VRAM creator GPU | RTX 4060 Ti 16GB, RTX 4080, RTX 5070 Ti, RTX 5080 | 14.5 GB usable VRAM | 28 GB | RAM offload | Open calculator |
| 24 GB VRAM homelab workstation | RTX 3090, RTX 4090 | 22.5 GB usable VRAM | 28 GB | RAM offload | Open calculator |
| 32 GB VRAM Blackwell workstation | RTX 5090 | 30.5 GB usable VRAM | 28 GB | RAM offload | Open calculator |
| 48 GB VRAM workstation | RTX A6000, L40S 48GB | 46.5 GB usable VRAM | 28 GB | Runs locally | Open calculator |
| Apple Silicon 32 GB unified memory | M2 Max 32GB, M3 Max 36GB | 26 GB unified | 28 GB | Too large | Open calculator |
| Apple Silicon 256 GB unified memory | Mac Studio M3 Ultra 256GB, Mac Studio M4 Ultra 256GB | 250 GB unified | 28 GB | Runs locally | Open calculator |
| Quantization | Estimated memory | Use case |
|---|---|---|
| Q4 / 4-bit | 28 GB | Default local inference balance |
This is a practical planning estimate, not a benchmark. Real memory use changes with backend, context length, KV cache, quantization file, drivers, and offloading settings.
35B · Agentic coding and multimodal reasoning when 27B is not enough and 32GB-class headroom is available
32B · Strong local coding and architecture work on 24GB GPUs
32B · Heavy local reasoning on 24GB GPUs
31B · High-quality multimodal reasoning, coding assistants, and local-first agent workflows