RunsOnMy ModelsGPUs & MacsBy VRAMAPI pricesVRAM calculatorNew

Can I run Llama 3.2 3B Instruct locally?

Llama 3.2 3B Instruct by meta-llama has 3.2B parameters. At 4-bit with an 8K context it needs about 3.4 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.

3.4 GBQ4_K_M · 8K context
RTX 4060smallest consumer GPU that fits
M1 16GBsmallest Mac that fits
128Kmax context
GQAattention type

Memory needed by quant and context length

GB of GPU memory = weights + KV cache (FP16) + runtime overhead. Weights use real file sizes from unsloth/Llama-3.2-3B-Instruct-GGUF where available (marked •), otherwise parameter count × bits per weight. How we calculate.

QuantQuality4K ctx8K ctx32K ctx128K ctx
2-bit (Q2_K) •rough2.4 GB2.8 GB5.4 GB15.9 GB
3-bit (Q3_K_M) •usable2.7 GB3.1 GB5.7 GB16.2 GB
4-bit (Q4_K_M) •good, most popular3.0 GB3.4 GB6.0 GB16.5 GB
5-bit (Q5_K_M) •very good3.2 GB3.7 GB6.3 GB16.8 GB
6-bit (Q6_K) •near lossless3.5 GB4.0 GB6.6 GB17.1 GB
8-bit (Q8_0) •effectively lossless4.3 GB4.7 GB7.3 GB17.8 GB
16-bit (F16/BF16) •original weights13.1 GB13.5 GB16.1 GB26.6 GB

Which GPUs and Macs run Llama 3.2 3B Instruct?

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).

HardwareMemoryFits?Best quantNeedsMax tok/s (est.)
RTX 3060 12GB12 GBFitsQ8_04.7 GB105details
RTX 4060 Ti 16GB16 GBFitsF1613.5 GB22details
RTX 309024 GBFitsF1613.5 GB73details
RTX 409024 GBFitsF1613.5 GB78details
RTX 509032 GBFitsF1613.5 GB139details
RTX 5070 Ti16 GBFitsF1613.5 GB70details
RX 7900 XTX24 GBFitsF1613.5 GB75details
Mac M4 Pro 48GB48 GBFitsF1613.5 GB21details
Mac M4 Max 128GB128 GBFitsF1613.5 GB42details
Mac M3 Ultra 512GB512 GBFitsF1613.5 GB64details
DGX Spark128 GBFitsF1613.5 GB21details
RTX 407012 GBFitsQ8_04.7 GB147details
RTX 5060 Ti 16GB16 GBFitsF1613.5 GB35details
RTX 40608 GBFitsQ8_04.7 GB79details
Arc B58012 GBFitsQ8_04.7 GB133details
RTX 3060 Ti8 GBFitsQ8_04.7 GB131details
RTX 30708 GBFitsQ8_04.7 GB131details
RTX 3080 10GB10 GBFitsQ8_04.7 GB222details
RTX 3080 12GB12 GBFitsQ8_04.7 GB267details
RTX 3090 Ti24 GBFitsF1613.5 GB78details
RTX 4060 Ti 8GB8 GBFitsQ8_04.7 GB84details
RTX 4070 Super12 GBFitsQ8_04.7 GB147details
RTX 4070 Ti12 GBFitsQ8_04.7 GB147details
RTX 4070 Ti Super16 GBFitsF1613.5 GB52details
RTX 408016 GBFitsF1613.5 GB56details
RTX 4080 Super16 GBFitsF1613.5 GB57details
RTX 50608 GBFitsQ8_04.7 GB131details
RTX 507012 GBFitsQ8_04.7 GB196details
RTX 508016 GBFitsF1613.5 GB75details
2x RTX 30902×24 GBFitsF1613.5 GB73details
2x RTX 40902×24 GBFitsF1613.5 GB78details
2x RTX 50902×32 GBFitsF1613.5 GB139details
RTX A600048 GBFitsF1613.5 GB60details
RTX 6000 Ada48 GBFitsF1613.5 GB75details
RTX PRO 6000 Blackwell96 GBFitsF1613.5 GB139details
L40S48 GBFitsF1613.5 GB67details
A100 40GB40 GBFitsF1613.5 GB121details
A100 80GB80 GBFitsF1613.5 GB158details
H100 PCIe80 GBFitsF1613.5 GB155details
H100 SXM80 GBFitsF1613.5 GB260details
H200141 GBFitsF1613.5 GB373details
RX 76008 GBFitsQ8_04.7 GB84details
RX 7700 XT12 GBFitsQ8_04.7 GB126details
RX 7800 XT16 GBFitsF1613.5 GB48details
RX 7900 GRE16 GBFitsF1613.5 GB45details
RX 7900 XT20 GBFitsF1613.5 GB62details
RX 907016 GBFitsF1613.5 GB50details
RX 9070 XT16 GBFitsF1613.5 GB50details
Ryzen AI Max+ 395 128GB128 GBFitsF1613.5 GB20details
Arc A770 16GB16 GBFitsF1613.5 GB44details
Mac M1 16GB16 GBFitsQ8_04.7 GB20details
Mac M2 24GB24 GBFitsF1613.5 GB7.8details
Mac M3 24GB24 GBFitsF1613.5 GB7.8details
Mac M4 16GB16 GBFitsQ8_04.7 GB35details
Mac M4 32GB32 GBFitsF1613.5 GB9.3details
Mac M1 Pro 32GB32 GBFitsF1613.5 GB16details
Mac M3 Pro 36GB36 GBFitsF1613.5 GB12details
Mac M4 Pro 24GB24 GBFitsF1613.5 GB21details
Mac M4 Pro 64GB64 GBFitsF1613.5 GB21details
Mac M1 Max 64GB64 GBFitsF1613.5 GB31details
Mac M2 Max 96GB96 GBFitsF1613.5 GB31details
Mac M3 Max 128GB128 GBFitsF1613.5 GB31details
Mac M4 Max 64GB64 GBFitsF1613.5 GB42details
Mac M2 Ultra 192GB192 GBFitsF1613.5 GB62details
Mac M3 Ultra 256GB256 GBFitsF1613.5 GB64details

Prefer an API?

Meta: Llama 3.2 3B Instruct is on OpenRouter at $0.0509 per 1M input tokens and $0.335 per 1M output tokens. Price details and history →

Questions

How much VRAM does Llama 3.2 3B Instruct need?

About 3.4 GB at 4-bit (Q4_K_M) with an 8K context: 1.9 GB of weights (real GGUF file size), 0.9 GB of KV cache and 0.6 GB of runtime overhead. At 8-bit it needs about 4.7 GB.

What is the smallest GPU that runs Llama 3.2 3B Instruct?

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 Llama 3.2 3B Instruct 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.

Llama 3.2 3B

3.2B · 3.6 GB at 4-bit

Llama 3.2 1B

1.2B · 1.8 GB at 4-bit

Llama 3.2 1B Instruct

1.2B · 1.6 GB at 4-bit

Hermes 3 Llama 3.2 3B

3.2B · 3.4 GB at 4-bit

granite 3.1 3b a800m instruct

3.3B · 2.9 GB at 4-bit

d1 3B

3.1B · 2.3 GB at 4-bit

Model facts

Hugging Face repometa-llama/Llama-3.2-3B-Instruct
Parameters3,212,749,824 (3.2B)
Layers / hidden size28 / 3072
AttentionGQA; 24 heads, 8 KV heads, head dim 128
KV cache per token (FP16)112.0 KiB
Max context131,072 tokens
Licensellama3.2 (gated: accept terms on Hugging Face)
Released on HFSep 18, 2024
Downloads (30d) / likes824,000 / 3,027

Sources: Hugging Face model API · config.json (unsloth/Llama-3.2-3B-Instruct-GGUF) · GGUF files · GPU specs from vendor pages (list). Data fetched Nov 17, 2026. Spot an error? Report it.