Local LLM model fit
gpt-oss-safeguard 20B is a 21B gpt-oss-safeguard model. This page estimates Q4 VRAM fit, Ollama command, context planning, and fallback choices for common local AI GPUs.
Local safety-policy classification, input-output filtering, moderation review, and trust-and-safety labeling on 16GB+ GPUs. Weakness: Purpose-built for safety classification, not general chat or coding; requires the Harmony response format and clear developer-provided policies.
| Hardware | Examples | Clean capacity | Q4 need | Status | Calculator |
|---|---|---|---|---|---|
| 6 GB VRAM entry GPU | GTX 1660, RTX 2060 6GB | 4.5 GB usable VRAM | 14 GB | Too large | Open calculator |
| 8 GB VRAM mainstream GPU | RTX 3060 Ti, RTX 4060, RTX 3070 | 6.5 GB usable VRAM | 14 GB | RAM offload | Open calculator |
| 10 GB VRAM older high-end GPU | RTX 3080 10GB | 8.5 GB usable VRAM | 14 GB | RAM offload | Open calculator |
| 12 GB VRAM local agent GPU | RTX 3060 12GB, RTX 4070, RTX 5070 | 10.5 GB usable VRAM | 14 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 | 14 GB | RAM offload | Open calculator |
| 24 GB VRAM homelab workstation | RTX 3090, RTX 4090 | 22.5 GB usable VRAM | 14 GB | Runs locally | Open calculator |
| 32 GB VRAM Blackwell workstation | RTX 5090 | 30.5 GB usable VRAM | 14 GB | Runs locally | Open calculator |
| 48 GB VRAM workstation | RTX A6000, L40S 48GB | 46.5 GB usable VRAM | 14 GB | Runs locally | Open calculator |
| Apple Silicon 32 GB unified memory | M2 Max 32GB, M3 Max 36GB | 26 GB unified | 14 GB | Runs locally | Open calculator |
| Apple Silicon 256 GB unified memory | Mac Studio M3 Ultra 256GB, Mac Studio M4 Ultra 256GB | 250 GB unified | 14 GB | Runs locally | Open calculator |
| Quantization | Estimated memory | Use case |
|---|---|---|
| Native publisher quantization | 14 GB | Only the published/runtime-native tag; precision is model-specific |
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.
20B · Local reasoning, agent planning, and tool-use workflows on 16GB+ GPUs
14B · Local coding, scripts, repo assistance, technical agents
14B · Local reasoning and debugging on 12GB/16GB GPUs
14B · Compact reasoning and technical assistant on 12GB/16GB GPUs