# Can I run gpt-oss 20B on Apple M4 (24GB)?

> **Yes, it runs.** Runs at Q4_K_M using ~13.2 GB of ~16 GB usable.

Last validated: 2026-06-15.

## Memory math
- gpt-oss 20B: 21B (MoE, 3.6B active); Q4_K_M weighs 11.28 GB on disk.
- At Q4_K_M with a 4k context it needs ~13.2 GB (weights + KV cache + ~0.8 GB runtime overhead, ±15% with context length).
- Apple M4 (24GB): 24 GB unified, ~16 GB usable for model weights.
- Headroom: ~2.8 GB.

## How to run
Recommended quant: Q4_K_M. Best tool on macOS: LM Studio.

```sh
ollama run gpt-oss:20b
```

## Run gpt-oss 20B on other devices
- [Nvidia GeForce RTX 4060 Ti (16GB)](https://localmodel.run/can-i-run/gpt-oss-20b/nvidia-rtx-4060-ti-16gb)
- [Nvidia GeForce RTX 4080 (16GB)](https://localmodel.run/can-i-run/gpt-oss-20b/nvidia-rtx-4080-16gb)
- [Apple M4 Pro (24GB)](https://localmodel.run/can-i-run/gpt-oss-20b/apple-m4-pro-24gb)
- [Nvidia GeForce RTX 4090 (24GB)](https://localmodel.run/can-i-run/gpt-oss-20b/nvidia-rtx-4090-24gb)
- [Nvidia GeForce RTX 3090 (24GB)](https://localmodel.run/can-i-run/gpt-oss-20b/nvidia-rtx-3090-24gb)
- [AMD Radeon RX 7900 XTX (24GB)](https://localmodel.run/can-i-run/gpt-oss-20b/amd-rx-7900-xtx-24gb)

## Other models that run on Apple M4 (24GB)
- [DeepSeek-V2-Lite (16B)](https://localmodel.run/can-i-run/deepseek-v2-lite/apple-m4-24gb)
- [Phi-4 14B (14B)](https://localmodel.run/can-i-run/phi-4-14b/apple-m4-24gb)
- [Qwen2.5 14B (14B)](https://localmodel.run/can-i-run/qwen2.5-14b/apple-m4-24gb)
- [Qwen3 14B (14B)](https://localmodel.run/can-i-run/qwen3-14b/apple-m4-24gb)
- [DeepSeek-R1-Distill-Qwen 14B (14B)](https://localmodel.run/can-i-run/deepseek-r1-distill-qwen-14b/apple-m4-24gb)
- [Qwen2.5 Coder 14B (14B)](https://localmodel.run/can-i-run/qwen2.5-coder-14b/apple-m4-24gb)

Estimate, not a guarantee. Sources: https://ollama.com/library/gpt-oss, https://ollama.com/library/gpt-oss/tags, https://huggingface.co/ggml-org/gpt-oss-20b-GGUF, https://huggingface.co/openai/gpt-oss-20b.
More: https://localmodel.run/can-i-run/gpt-oss-20b/apple-m4-24gb