Can Desktop PC · GeForce RTX 5070 · 32 GB RAM run gpt-oss-20b?
12 GB VRAM + 32 GB RAM · 672 GB/s memory bandwidth · gpt-oss-20b has 20.9B parameters (3.6B active per token).
Short answer
It loads, but it's borderline even for chat.
Best quantization for chat: MXFP4 (11 GB of weights) on Ollama. Generation: 60–100 tok/s, leaving about 0.0 GB free for your other apps.
How it rates for each workload
Same model and machine, different jobs. Agents and long documents send far bigger prompts than chat, so they need much more speed and memory.
| Workload | Verdict | Quant | Generation | First reply | Context | |
|---|---|---|---|---|---|---|
| Chatwith Open WebUI | Borderline | MXFP4 | 60–100 tok/s | under 1 s | 8K | Details |
| Coding questionswith Continue | Borderline | MXFP4 | 50–85 tok/s | ≈2 s | 16K | Details |
| Coding in a repositorywith Aider | Borderline | MXFP4 | 40–70 tok/s | ≈7 s | 32K | Details |
| Agentic codingwith OpenCode | Borderline | MXFP4 | 40–65 tok/s | ≈10 s | 32K | Details |
| Long documents (64K)with Open WebUI | Borderline | MXFP4 | 25–40 tok/s | ≈35 s | 64K | Details |
Speeds are bandwidth-based estimates; no direct benchmark exists for this pair. “First reply” is the time to read the workload’s opening prompt. How we calculate
Get it running
Copy-paste commands for the MXFP4 build with a 8K context. Ollama and llama.cpp download the file straight from Hugging Face.
ollama run hf.co/unsloth/gpt-oss-20b-GGUF:F16
Downloads 13.79 GB (F16). Ollama's default context is small: set OLLAMA_CONTEXT_LENGTH=8192 before `ollama serve` for long prompts.
Other models for Desktop PC · GeForce RTX 5070 · 32 GB RAM
The best-rated models for chat on this machine.
- MiniCPM5 2B120–200 tok/sExcellent
- Qwen3.5 9B40–65 tok/sExcellent
- Qwen3.5 4B75–120 tok/sExcellent
- Gemma 4 12B40–65 tok/sExcellent
- GLM-4.6V-Flash (10B)40–65 tok/sExcellent
- Qwen3 8B45–75 tok/sExcellent
Cheapest machines for gpt-oss-20b
Hardware rated comfortable or better for chat, lowest price first.
- Mac mini M4 24 GB$999 · 25–35 tok/sExcellent
- Mac mini M6 24 GB$1,299 · 30–50 tok/sExcellent
- Laptop · Ryzen AI 9 HX 370 (Radeon 890M) · 32 GB$1,300 · 20–35 tok/sExcellent
- Laptop · Core Ultra 7 258V (Arc 140V) · 32 GB$1,300 · 18–30 tok/sExcellent
- Desktop PC · GeForce RTX 5060 Ti 16GB · 32 GB RAM$1,330 · ≈100 tok/sExcellent
- Desktop PC · GeForce RTX 4060 Ti 16GB · 32 GB RAM$1,350 · 50–85 tok/sExcellent
Different tool, context size or workload?
These pages use typical settings. The full check models your exact setup.