Can my computer run gpt-oss 20B?
OpenAI's open model: thinks before it answers and uses tools, fast for its size. gpt-oss 20B needs about 13.1 GB of memory. It runs comfortably on Macs with 16 GB or more, and on PCs with 16 GB of RAM or more, even without a graphics card.
gpt-oss 20B on common computers
Graded with Toskar 1.14.0's own estimate: the model file plus working memory at an 8K context.
| Computer | Fit | What to expect |
|---|---|---|
| Mac, 8 GB | Too big | Needs more memory than this computer has. |
| Mac, 16 GB | Good fit | Should run comfortably. |
| Mac, 24 GB | Excellent fit | Plenty of memory to spare. |
| Mac, 32 GB | Excellent fit | Plenty of memory to spare. |
| Mac, 48 GB | Excellent fit | Plenty of memory to spare. |
| Mac, 64 GB | Excellent fit | Plenty of memory to spare. |
| Mac, 128 GB | Excellent fit | Plenty of memory to spare. |
| PC, 8 GB, no graphics card | Too big | Needs more memory than this computer has. |
| PC, 16 GB, no graphics card | Good fit | Should run comfortably. |
| PC, 32 GB, no graphics card | Excellent fit | Plenty of memory to spare. |
| PC, 16 GB + 8 GB graphics card | Good fit | Should run comfortably. Partial GPU offload expected. |
| PC, 32 GB + 12 GB graphics card | Excellent fit | Plenty of memory to spare. Partial GPU offload expected. |
| PC, 32 GB + 16 GB graphics card | Excellent fit | Plenty of memory to spare. Good GPU fit. |
| PC, 64 GB + 24 GB graphics card | Excellent fit | Plenty of memory to spare. Good GPU fit. |
About gpt-oss 20B
- Memory needed
- ≈ 13.1 GB
- Download size
- 11.3 GB
- Parameters
- 21B
- Quantization
- MXFP4
- Longest conversation
- 128K tokens
Good at
- Chat and writing
- Coding
- Step-by-step reasoning
- Uses tools, such as web search and your files
Languages it handles well
English
Similar models to compare
- Qwen 2.5 14B≈ 10.2 GB
Largest curated quality pick when your machine has headroom.
- Qwen 2.5 Coder 14B≈ 9.7 GB
Excellent programming model for larger refactors and multi-file work.
- Muse Glimmer 30B≈ 17.1 GB
Meta's large model that sees pictures, codes, and reasons well; for 32 GB Macs or 24 GB graphics cards.
- Qwen 3.5 9B≈ 7.2 GB
Strong everyday assistant that also sees pictures.
Questions
- How much memory does gpt-oss 20B need?
- About 13.1 GB with Toskar's default 8K context: the 11.3 GB model file plus room for the conversation. After it has run once, Toskar measures what it really uses on your computer.
- Can gpt-oss 20B run on a Mac with 8 GB?
- No. It needs about 13.1 GB, more than a Mac with 8 GB has. A smaller model, or a paired computer with more memory, can answer instead.
- Can gpt-oss 20B run without a graphics card?
- Yes. gpt-oss 20B runs comfortably on a PC with 16 GB of RAM and no graphics card, using about 13.1 GB. Without a graphics card it runs on the processor, so it answers more slowly than with one. On a Mac with Apple silicon, the graphics share the Mac's memory, so no separate card is needed.
- Is gpt-oss 20B free to use?
- Yes. Toskar Core is free and open source, and it downloads gpt-oss 20B for you at no cost. Everything runs on your computer, so there's no per-question bill.
Every model
- Llama 3.2 1B≈ 1.3 GB
- Qwen 2.5 Coder 1.5B≈ 1.6 GB
- DeepSeek R1 Distill 1.5B≈ 1.6 GB
- Gemma 2 2B≈ 2.2 GB
- Qwen 2.5 3B≈ 2.5 GB
- Llama 3.2 3B≈ 2.6 GB
- Phi 3.5 Mini≈ 3.0 GB
- Gemma 3 4B≈ 3.9 GB
- Qwen 3.5 4B≈ 3.9 GB
- Mistral 7B Instruct≈ 5.0 GB
- Qwen 2.5 Coder 7B≈ 5.3 GB
- Qwen 2.5 7B≈ 5.3 GB
- DeepSeek R1 Distill 7B≈ 5.3 GB
- Qwen 2.5 VL 7B≈ 6.1 GB
- Gemma 2 9B≈ 6.5 GB
- Qwen 3.5 9B≈ 7.2 GB
- Qwen 2.5 Coder 14B≈ 9.7 GB
- Qwen 2.5 14B≈ 10.2 GB
- gpt-oss 20B≈ 13.1 GB
- Muse Glimmer 30B≈ 17.1 GB
- Qwen 3.8 27B≈ 20.4 GB
- Qwen 2.5 32B≈ 21.5 GB
- Qwen 3.6 35B-A3B≈ 21.8 GB
- gpt-oss 120B≈ 66.7 GB
Run gpt-oss 20B with Toskar
Toskar Core is free and open source for Mac, Windows, and Linux. It checks what your computer can run, downloads the model for you, and keeps every question on your computer.