Can I run Phi 2 locally?
Phi 2 by microsoft has 2.8B parameters. At 4-bit with an 8K context it needs about 4.7 GB of GPU memory. It fits on 65 of 65 GPUs and Macs we track. See all models that fit in 8 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 TheBloke/phi-2-GGUF where available (marked •), otherwise parameter count × bits per weight. How we calculate.
| Quant | Quality |
|---|---|
| 2-bit (Q2_K) • | rough |
| 3-bit (Q3_K_M) • | usable |
| 4-bit (Q4_K_M) • | good, most popular |
| 5-bit (Q5_K_M) • | very good |
| 6-bit (Q6_K) • | near lossless |
| 8-bit (Q8_0) • | effectively lossless |
| 16-bit (F16/BF16) | original weights |
Which GPUs and Macs run Phi 2?
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 | Fits | F16 | 8.2 GB | 65 | details |
| RTX 4060 Ti 16GB | 16 GB | Fits | F16 | 8.2 GB | 52 | details |
| RTX 3090 | 24 GB | Fits | F16 | 8.2 GB | 168 | details |
| RTX 4090 | 24 GB | Fits | F16 | 8.2 GB | 181 | details |
| RTX 5090 | 32 GB | Fits | F16 | 8.2 GB | 322 | details |
| RTX 5070 Ti | 16 GB | Fits | F16 | 8.2 GB | 161 | details |
| RX 7900 XTX | 24 GB | Fits | F16 | 8.2 GB | 173 | details |
| Mac M4 Pro 48GB | 48 GB | Fits | F16 | 8.2 GB | 49 | details |
| Mac M4 Max 128GB | 128 GB | Fits | F16 | 8.2 GB | 98 | details |
| Mac M3 Ultra 512GB | 512 GB | Fits | F16 | 8.2 GB | 147 | details |
| DGX Spark | 128 GB | Fits | F16 | 8.2 GB | 49 | details |
| RTX 4070 | 12 GB | Fits | F16 | 8.2 GB | 91 | details |
| RTX 5060 Ti 16GB | 16 GB | Fits | F16 | 8.2 GB | 81 | details |
| RTX 4060 | 8 GB | Fits | Q8_0 | 5.8 GB | 92 | details |
| Arc B580 | 12 GB | Fits | F16 | 8.2 GB | 82 | details |
| RTX 3060 Ti | 8 GB | Fits | Q8_0 | 5.8 GB | 151 | details |
| RTX 3070 | 8 GB | Fits | Q8_0 | 5.8 GB | 151 | details |
| RTX 3080 10GB | 10 GB | Fits | F16 | 8.2 GB | 137 | details |
| RTX 3080 12GB | 12 GB | Fits | F16 | 8.2 GB | 164 | details |
| RTX 3090 Ti | 24 GB | Fits | F16 | 8.2 GB | 181 | details |
| RTX 4060 Ti 8GB | 8 GB | Fits | Q8_0 | 5.8 GB | 97 | details |
| RTX 4070 Super | 12 GB | Fits | F16 | 8.2 GB | 91 | details |
| RTX 4070 Ti | 12 GB | Fits | F16 | 8.2 GB | 91 | details |
| RTX 4070 Ti Super | 16 GB | Fits | F16 | 8.2 GB | 121 | details |
| RTX 4080 | 16 GB | Fits | F16 | 8.2 GB | 129 | details |
| RTX 4080 Super | 16 GB | Fits | F16 | 8.2 GB | 132 | details |
| RTX 5060 | 8 GB | Fits | Q8_0 | 5.8 GB | 151 | details |
| RTX 5070 | 12 GB | Fits | F16 | 8.2 GB | 121 | details |
| RTX 5080 | 16 GB | Fits | F16 | 8.2 GB | 173 | details |
| 2x RTX 3090 | 2×24 GB | Fits | F16 | 8.2 GB | 168 | details |
| 2x RTX 4090 | 2×24 GB | Fits | F16 | 8.2 GB | 181 | details |
| 2x RTX 5090 | 2×32 GB | Fits | F16 | 8.2 GB | 322 | details |
| RTX A6000 | 48 GB | Fits | F16 | 8.2 GB | 138 | details |
| RTX 6000 Ada | 48 GB | Fits | F16 | 8.2 GB | 173 | details |
| RTX PRO 6000 Blackwell | 96 GB | Fits | F16 | 8.2 GB | 322 | details |
| L40S | 48 GB | Fits | F16 | 8.2 GB | 155 | details |
| A100 40GB | 40 GB | Fits | F16 | 8.2 GB | 280 | details |
| A100 80GB | 80 GB | Fits | F16 | 8.2 GB | 367 | details |
| H100 PCIe | 80 GB | Fits | F16 | 8.2 GB | 360 | details |
| H100 SXM | 80 GB | Fits | F16 | 8.2 GB | 603 | details |
| H200 | 141 GB | Fits | F16 | 8.2 GB | 863 | details |
| RX 7600 | 8 GB | Fits | Q8_0 | 5.8 GB | 97 | details |
| RX 7700 XT | 12 GB | Fits | F16 | 8.2 GB | 78 | details |
| RX 7800 XT | 16 GB | Fits | F16 | 8.2 GB | 112 | details |
| RX 7900 GRE | 16 GB | Fits | F16 | 8.2 GB | 104 | details |
| RX 7900 XT | 20 GB | Fits | F16 | 8.2 GB | 144 | details |
| RX 9070 | 16 GB | Fits | F16 | 8.2 GB | 115 | details |
| RX 9070 XT | 16 GB | Fits | F16 | 8.2 GB | 115 | details |
| Ryzen AI Max+ 395 128GB | 128 GB | Fits | F16 | 8.2 GB | 46 | details |
| Arc A770 16GB | 16 GB | Fits | F16 | 8.2 GB | 101 | details |
| Mac M1 16GB | 16 GB | Fits | F16 | 8.2 GB | 12 | details |
| Mac M2 24GB | 24 GB | Fits | F16 | 8.2 GB | 18 | details |
| Mac M3 24GB | 24 GB | Fits | F16 | 8.2 GB | 18 | details |
| Mac M4 16GB | 16 GB | Fits | F16 | 8.2 GB | 22 | details |
| Mac M4 32GB | 32 GB | Fits | F16 | 8.2 GB | 22 | details |
| Mac M1 Pro 32GB | 32 GB | Fits | F16 | 8.2 GB | 36 | details |
| Mac M3 Pro 36GB | 36 GB | Fits | F16 | 8.2 GB | 27 | details |
| Mac M4 Pro 24GB | 24 GB | Fits | F16 | 8.2 GB | 49 | details |
| Mac M4 Pro 64GB | 64 GB | Fits | F16 | 8.2 GB | 49 | details |
| Mac M1 Max 64GB | 64 GB | Fits | F16 | 8.2 GB | 72 | details |
| Mac M2 Max 96GB | 96 GB | Fits | F16 | 8.2 GB | 72 | details |
| Mac M3 Max 128GB | 128 GB | Fits | F16 | 8.2 GB | 72 | details |
| Mac M4 Max 64GB | 64 GB | Fits | F16 | 8.2 GB | 98 | details |
| Mac M2 Ultra 192GB | 192 GB | Fits | F16 | 8.2 GB | 144 | details |
| Mac M3 Ultra 256GB | 256 GB | Fits | F16 | 8.2 GB | 147 | details |
Questions
How much VRAM does Phi 2 need?
About 4.7 GB at 4-bit (Q4_K_M) with an 8K context: 1.7 GB of weights (real GGUF file size), 2.5 GB of KV cache and 0.5 GB of runtime overhead. At 8-bit it needs about 5.8 GB.
What is the smallest GPU that runs Phi 2?
At Q4_K_M and 8K context, the smallest single consumer GPU in our table that fits it is the RTX 4060 (8 GB).
Can I run Phi 2 on a Mac?
Yes. At Q4_K_M with 8K context it fits on a Mac M1 16GB 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.
Phi 3 Mini 4K Instruct
3.8B · 5.7 GB at 4-bitPhi 3.5 Mini Instruct
3.8B · 5.7 GB at 4-bitPhi 3 Mini 128K Instruct
3.8B · 5.7 GB at 4-bitLFM2.5 2.6B
2.7B · 2.3 GB at 4-bitGranite 3.3 2B Instruct
2.5B · 2.6 GB at 4-bitMiniCPM5 2B
2.5B · 2.5 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/phi-2/) Widget (script)
<div class="runsonmy-widget" data-model="phi-2"><a href="https://runsonmy.com/model/phi-2/">Will Phi 2 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/phi-2/" width="100%" height="220" style="max-width:440px;border:0" loading="lazy" title="Will Phi 2 run on my GPU? (RunsOnMy)"></iframe> Model facts
| Hugging Face repo | microsoft/phi-2 |
| Parameters | 2,779,683,840 (2.8B) |
| Layers / hidden size | 32 / 2560 |
| Attention | MHA; 32 heads, 32 KV heads, head dim 80 |
| KV cache per token (FP16) | 320.0 KiB |
| Max context | 2,048 tokens |
| License | mit |
| Released on HF | Dec 13, 2023 |
| Downloads (30d) / likes | 177,357 / 3,537 |
Sources: Hugging Face model API · config.json (microsoft/phi-2) · GGUF files · GPU specs from vendor pages (list). Data fetched Nov 17, 2026. Spot an error? Report it.