Can I run gpt-oss 20B locally?
gpt-oss 20B by openai has 21B parameters (4.2B active per token, 32 experts, 4 used per token). At 4-bit with an 8K context it needs about 11.8 GB of GPU memory. It fits on 48 of 65 GPUs and Macs we track. See all models that fit in 16 GB of VRAM.
Memory needed by quant and context length
GB of GPU memory = weights + KV cache (FP16) + runtime overhead. Weights use real file sizes from bartowski/openai_gpt-oss-20b-GGUF where available (marked •), otherwise parameter count × bits per weight. How we calculate.
| Quant | Quality | 4K ctx | 8K ctx | 32K ctx | 128K ctx |
|---|---|---|---|---|---|
| 2-bit (Q2_K) • | rough | 11.6 GB | 11.7 GB | 12.3 GB | 14.6 GB |
| 3-bit (Q3_K_M) • | usable | 11.6 GB | 11.7 GB | 12.3 GB | 14.6 GB |
| 4-bit (Q4_K_M) • | good, most popular | 11.8 GB | 11.8 GB | 12.4 GB | 14.7 GB |
| 5-bit (Q5_K_M) • | very good | 11.8 GB | 11.9 GB | 12.5 GB | 14.7 GB |
| 6-bit (Q6_K) • | near lossless | 12.1 GB | 12.2 GB | 12.8 GB | 15.0 GB |
| 8-bit (Q8_0) • | effectively lossless | 12.2 GB | 12.3 GB | 12.8 GB | 15.1 GB |
| 16-bit (F16/BF16) • | original weights | 13.7 GB | 13.8 GB | 14.4 GB | 16.6 GB |
gpt-oss 20B on Ollama: download size and VRAM
ollama run gpt-oss:20b | Source | File | Download size | GPU memory at 8K ctx (est.) |
|---|---|---|---|
| Ollama download size | gpt-oss:20b (MXFP4) | 14 GB | 14.0 GB |
| Our GGUF (Hugging Face) | bartowski/openai_gpt-oss-20b-GGUF (Q4_K_M) | 11.7 GB | 11.8 GB |
Download sizes are decimal GB, as ollama.com lists them; memory figures on this site use binary GB (GiB). GPU memory for the Ollama tag is its file size plus this model's KV cache at 8K context and runtime overhead; Ollama's default context window may be smaller (set num_ctx to change it). Source: ollama.com/library/gpt-oss:20b, checked Nov 17, 2026. Refreshed weekly; shown only when the tag's architecture and parameter count match this model. All Ollama models with sizes and VRAM.
Which GPUs and Macs run gpt-oss 20B?
Short answer with CPU offload and RAM needed: Can I run gpt-oss-20b on my PC?
Best quality quant that fits at 8K context. "Tight" = within 10% of usable memory. Speed is a ceiling from memory bandwidth (real speed is typically 50–80% of it).
| Hardware | Memory | Fits? | Best quant | Needs | Max tok/s (est.) | Details |
|---|---|---|---|---|---|---|
| RTX 3060 12GB | 12 GB | No | — | — | — | details |
| RTX 4060 Ti 16GB | 16 GB | Fits | Q8_0 | 12.3 GB | 119 | details |
| RTX 3090 | 24 GB | Fits | F16 | 13.8 GB | 338 | details |
| RTX 4090 | 24 GB | Fits | F16 | 13.8 GB | 365 | details |
| RTX 5090 | 32 GB | Fits | F16 | 13.8 GB | 648 | details |
| RTX 5070 Ti | 16 GB | Fits | Q8_0 | 12.3 GB | 369 | details |
| RX 7900 XTX | 24 GB | Fits | F16 | 13.8 GB | 347 | details |
| Mac M4 Pro 48GB | 48 GB | Fits | F16 | 13.8 GB | 99 | details |
| Mac M4 Max 128GB | 128 GB | Fits | F16 | 13.8 GB | 197 | details |
| Mac M3 Ultra 512GB | 512 GB | Fits | F16 | 13.8 GB | 296 | details |
| DGX Spark | 128 GB | Fits | F16 | 13.8 GB | 99 | details |
| RTX 4070 | 12 GB | No | — | — | — | details |
| RTX 5060 Ti 16GB | 16 GB | Fits | Q8_0 | 12.3 GB | 185 | details |
| RTX 4060 | 8 GB | No | — | — | — | details |
| Arc B580 | 12 GB | No | — | — | — | details |
| RTX 3060 Ti | 8 GB | No | — | — | — | details |
| RTX 3070 | 8 GB | No | — | — | — | details |
| RTX 3080 10GB | 10 GB | No | — | — | — | details |
| RTX 3080 12GB | 12 GB | No | — | — | — | details |
| RTX 3090 Ti | 24 GB | Fits | F16 | 13.8 GB | 365 | details |
| RTX 4060 Ti 8GB | 8 GB | No | — | — | — | details |
| RTX 4070 Super | 12 GB | No | — | — | — | details |
| RTX 4070 Ti | 12 GB | No | — | — | — | details |
| RTX 4070 Ti Super | 16 GB | Fits | Q8_0 | 12.3 GB | 277 | details |
| RTX 4080 | 16 GB | Fits | Q8_0 | 12.3 GB | 295 | details |
| RTX 4080 Super | 16 GB | Fits | Q8_0 | 12.3 GB | 303 | details |
| RTX 5060 | 8 GB | No | — | — | — | details |
| RTX 5070 | 12 GB | No | — | — | — | details |
| RTX 5080 | 16 GB | Fits | Q8_0 | 12.3 GB | 395 | details |
| 2x RTX 3090 | 2×24 GB | Fits | F16 | 13.8 GB | 338 | details |
| 2x RTX 4090 | 2×24 GB | Fits | F16 | 13.8 GB | 365 | details |
| 2x RTX 5090 | 2×32 GB | Fits | F16 | 13.8 GB | 648 | details |
| RTX A6000 | 48 GB | Fits | F16 | 13.8 GB | 278 | details |
| RTX 6000 Ada | 48 GB | Fits | F16 | 13.8 GB | 347 | details |
| RTX PRO 6000 Blackwell | 96 GB | Fits | F16 | 13.8 GB | 648 | details |
| L40S | 48 GB | Fits | F16 | 13.8 GB | 312 | details |
| A100 40GB | 40 GB | Fits | F16 | 13.8 GB | 562 | details |
| A100 80GB | 80 GB | Fits | F16 | 13.8 GB | 737 | details |
| H100 PCIe | 80 GB | Fits | F16 | 13.8 GB | 723 | details |
| H100 SXM | 80 GB | Fits | F16 | 13.8 GB | 1211 | details |
| H200 | 141 GB | Fits | F16 | 13.8 GB | 1736 | details |
| RX 7600 | 8 GB | No | — | — | — | details |
| RX 7700 XT | 12 GB | No | — | — | — | details |
| RX 7800 XT | 16 GB | Fits | Q8_0 | 12.3 GB | 257 | details |
| RX 7900 GRE | 16 GB | Fits | Q8_0 | 12.3 GB | 237 | details |
| RX 7900 XT | 20 GB | Fits | F16 | 13.8 GB | 289 | details |
| RX 9070 | 16 GB | Fits | Q8_0 | 12.3 GB | 264 | details |
| RX 9070 XT | 16 GB | Fits | Q8_0 | 12.3 GB | 264 | details |
| Ryzen AI Max+ 395 128GB | 128 GB | Fits | F16 | 13.8 GB | 93 | details |
| Arc A770 16GB | 16 GB | Fits | Q8_0 | 12.3 GB | 231 | details |
| Mac M1 16GB | 16 GB | No | — | — | — | details |
| Mac M2 24GB | 24 GB | Fits | F16 | 13.8 GB | 36 | details |
| Mac M3 24GB | 24 GB | Fits | F16 | 13.8 GB | 36 | details |
| Mac M4 16GB | 16 GB | No | — | — | — | details |
| Mac M4 32GB | 32 GB | Fits | F16 | 13.8 GB | 43 | details |
| Mac M1 Pro 32GB | 32 GB | Fits | F16 | 13.8 GB | 72 | details |
| Mac M3 Pro 36GB | 36 GB | Fits | F16 | 13.8 GB | 54 | details |
| Mac M4 Pro 24GB | 24 GB | Fits | F16 | 13.8 GB | 99 | details |
| Mac M4 Pro 64GB | 64 GB | Fits | F16 | 13.8 GB | 99 | details |
| Mac M1 Max 64GB | 64 GB | Fits | F16 | 13.8 GB | 145 | details |
| Mac M2 Max 96GB | 96 GB | Fits | F16 | 13.8 GB | 145 | details |
| Mac M3 Max 128GB | 128 GB | Fits | F16 | 13.8 GB | 145 | details |
| Mac M4 Max 64GB | 64 GB | Fits | F16 | 13.8 GB | 197 | details |
| Mac M2 Ultra 192GB | 192 GB | Fits | F16 | 13.8 GB | 289 | details |
| Mac M3 Ultra 256GB | 256 GB | Fits | F16 | 13.8 GB | 296 | details |
Prefer an API?
OpenAI: gpt-oss-20b is on OpenRouter at $0.018 per 1M input tokens and $0.09 per 1M output tokens. Price details and history →
Questions
How much VRAM does gpt-oss 20B need?
About 11.8 GB at 4-bit (Q4_K_M) with an 8K context: 10.9 GB of weights (real GGUF file size), 0.2 GB of KV cache and 0.8 GB of runtime overhead. At 8-bit it needs about 12.3 GB.
What is the smallest GPU that runs gpt-oss 20B?
At Q4_K_M and 8K context, the smallest single consumer GPU in our table that fits it is the RTX 4060 Ti 16GB (16 GB).
How big is the Ollama download for gpt-oss 20B?
14GB for the gpt-oss:20b tag (MXFP4), as listed on ollama.com. Run it with "ollama run gpt-oss:20b". With an 8K context, plan on about 14.0 GB of GPU memory.
Can I run gpt-oss 20B on a Mac?
Yes. At Q4_K_M with 8K context it fits on a Mac M2 24GB or larger, using the default macOS GPU memory limit.
Similar models
Same family or a similar parameter count, with the VRAM each needs at Q4_K_M and 8K context.
gpt-oss Safeguard 20B
22B · 11.8 GB at 4-bitgpt-oss 120B
117B · 59.6 GB at 4-bitgpt-oss Safeguard 120B
117B · 59.5 GB at 4-bitERNIE 4.5 21B A3B Thinking
22B · 13.6 GB at 4-bitERNIE 4.5 21B A3B PT
22B · 13.6 GB at 4-bitLFM2.5 24B A2B
24B · 14.3 GB at 4-bitEmbed this
Free badge and "Will it run on my GPU?" widget for model cards, READMEs and blogs. No cookies, no tracking. All options.
Copy a snippet
Markdown badge
[](https://runsonmy.com/model/gpt-oss-20b/) Widget (script)
<div class="runsonmy-widget" data-model="gpt-oss-20b"><a href="https://runsonmy.com/model/gpt-oss-20b/">Will gpt-oss 20B run on my GPU? Check on RunsOnMy</a></div>
<script async src="https://runsonmy.com/embed.js"></script> Widget (iframe)
<iframe src="https://runsonmy.com/embed/m/gpt-oss-20b/" width="100%" height="220" style="max-width:440px;border:0" loading="lazy" title="Will gpt-oss 20B run on my GPU? (RunsOnMy)"></iframe> Model facts
| Hugging Face repo | openai/gpt-oss-20b |
| Parameters | 20,914,757,184 (21B) · ~4.2B active (computed from config) |
| Layers / hidden size | 24 / 2880 |
| Attention | GQA + sliding window; 64 heads, 8 KV heads, head dim 64; sliding window 128 |
| KV cache per token (FP16) | 48.0 KiB (sliding layers capped at the window) |
| Max context | 131,072 tokens |
| License | apache-2.0 |
| Released on HF | Aug 4, 2025 |
| Downloads (30d) / likes | 5,596,946 / 5,325 |
Sources: Hugging Face model API · config.json (openai/gpt-oss-20b) · GGUF files · GPU specs from vendor pages (list). Data fetched Nov 17, 2026. Spot an error? Report it.