RunsOnMy ModelsGPUs & MacsBy VRAMAPI pricesVRAM calculatorNew

Can I run granite 3.3 2b instruct locally?

granite 3.3 2b instruct by ibm-granite has 2.5B parameters. At 4-bit with an 8K context it needs about 2.6 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.

2.6 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 ibm-granite/granite-3.3-2b-instruct-GGUF where available (marked •), otherwise parameter count × bits per weight. How we calculate.

QuantQuality4K ctx8K ctx32K ctx128K ctx
2-bit (Q2_K) •rough1.7 GB2.0 GB3.9 GB11.4 GB
3-bit (Q3_K_M) •usable2.0 GB2.3 GB4.2 GB11.7 GB
4-bit (Q4_K_M) •good, most popular2.2 GB2.6 GB4.4 GB11.9 GB
5-bit (Q5_K_M) •very good2.5 GB2.8 GB4.7 GB12.2 GB
6-bit (Q6_K) •near lossless2.7 GB3.1 GB4.9 GB12.4 GB
8-bit (Q8_0) •effectively lossless3.3 GB3.6 GB5.5 GB13.0 GB
16-bit (F16/BF16) •original weights5.5 GB5.8 GB7.7 GB15.2 GB

Which GPUs and Macs run granite 3.3 2b 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 GBFitsF165.8 GB71details
RTX 4060 Ti 16GB16 GBFitsF165.8 GB57details
RTX 309024 GBFitsF165.8 GB185details
RTX 409024 GBFitsF165.8 GB199details
RTX 509032 GBFitsF165.8 GB354details
RTX 5070 Ti16 GBFitsF165.8 GB177details
RX 7900 XTX24 GBFitsF165.8 GB189details
Mac M4 Pro 48GB48 GBFitsF165.8 GB54details
Mac M4 Max 128GB128 GBFitsF165.8 GB108details
Mac M3 Ultra 512GB512 GBFitsF165.8 GB162details
DGX Spark128 GBFitsF165.8 GB54details
RTX 407012 GBFitsF165.8 GB99details
RTX 5060 Ti 16GB16 GBFitsF165.8 GB88details
RTX 40608 GBFitsF165.8 GB54details
Arc B58012 GBFitsF165.8 GB90details
RTX 3060 Ti8 GBFitsF165.8 GB88details
RTX 30708 GBFitsF165.8 GB88details
RTX 3080 10GB10 GBFitsF165.8 GB150details
RTX 3080 12GB12 GBFitsF165.8 GB180details
RTX 3090 Ti24 GBFitsF165.8 GB199details
RTX 4060 Ti 8GB8 GBFitsF165.8 GB57details
RTX 4070 Super12 GBFitsF165.8 GB99details
RTX 4070 Ti12 GBFitsF165.8 GB99details
RTX 4070 Ti Super16 GBFitsF165.8 GB133details
RTX 408016 GBFitsF165.8 GB141details
RTX 4080 Super16 GBFitsF165.8 GB145details
RTX 50608 GBFitsF165.8 GB88details
RTX 507012 GBFitsF165.8 GB133details
RTX 508016 GBFitsF165.8 GB189details
2x RTX 30902×24 GBFitsF165.8 GB185details
2x RTX 40902×24 GBFitsF165.8 GB199details
2x RTX 50902×32 GBFitsF165.8 GB354details
RTX A600048 GBFitsF165.8 GB152details
RTX 6000 Ada48 GBFitsF165.8 GB189details
RTX PRO 6000 Blackwell96 GBFitsF165.8 GB354details
L40S48 GBFitsF165.8 GB170details
A100 40GB40 GBFitsF165.8 GB307details
A100 80GB80 GBFitsF165.8 GB402details
H100 PCIe80 GBFitsF165.8 GB395details
H100 SXM80 GBFitsF165.8 GB661details
H200141 GBFitsF165.8 GB947details
RX 76008 GBFitsF165.8 GB57details
RX 7700 XT12 GBFitsF165.8 GB85details
RX 7800 XT16 GBFitsF165.8 GB123details
RX 7900 GRE16 GBFitsF165.8 GB114details
RX 7900 XT20 GBFitsF165.8 GB158details
RX 907016 GBFitsF165.8 GB126details
RX 9070 XT16 GBFitsF165.8 GB126details
Ryzen AI Max+ 395 128GB128 GBFitsF165.8 GB51details
Arc A770 16GB16 GBFitsF165.8 GB110details
Mac M1 16GB16 GBFitsF165.8 GB13details
Mac M2 24GB24 GBFitsF165.8 GB20details
Mac M3 24GB24 GBFitsF165.8 GB20details
Mac M4 16GB16 GBFitsF165.8 GB24details
Mac M4 32GB32 GBFitsF165.8 GB24details
Mac M1 Pro 32GB32 GBFitsF165.8 GB39details
Mac M3 Pro 36GB36 GBFitsF165.8 GB30details
Mac M4 Pro 24GB24 GBFitsF165.8 GB54details
Mac M4 Pro 64GB64 GBFitsF165.8 GB54details
Mac M1 Max 64GB64 GBFitsF165.8 GB79details
Mac M2 Max 96GB96 GBFitsF165.8 GB79details
Mac M3 Max 128GB128 GBFitsF165.8 GB79details
Mac M4 Max 64GB64 GBFitsF165.8 GB108details
Mac M2 Ultra 192GB192 GBFitsF165.8 GB158details
Mac M3 Ultra 256GB256 GBFitsF165.8 GB162details

Questions

How much VRAM does granite 3.3 2b instruct need?

About 2.6 GB at 4-bit (Q4_K_M) with an 8K context: 1.4 GB of weights (real GGUF file size), 0.6 GB of KV cache and 0.5 GB of runtime overhead. At 8-bit it needs about 3.6 GB.

What is the smallest GPU that runs granite 3.3 2b 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 granite 3.3 2b 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.

granite 3.1 3b a800m instruct

3.3B · 2.9 GB at 4-bit

granite 4.1 3b

3.4B · 3.2 GB at 4-bit

granite 4.0 micro

3.4B · 3.2 GB at 4-bit

MiniCPM5 2B

2.5B · 2.5 GB at 4-bit

WeVisOne VL 2B

2.4B · 2.8 GB at 4-bit

LFM2.5 2.6B

2.7B · 2.3 GB at 4-bit

Model facts

Hugging Face repoibm-granite/granite-3.3-2b-instruct
Parameters2,533,539,840 (2.5B)
Layers / hidden size40 / 2048
AttentionGQA; 32 heads, 8 KV heads, head dim 64
KV cache per token (FP16)80.0 KiB
Max context131,072 tokens
Licenseapache-2.0
Released on HFApr 9, 2025
Downloads (30d) / likes40,713 / 88

Sources: Hugging Face model API · config.json (ibm-granite/granite-3.3-2b-instruct) · GGUF files · GPU specs from vendor pages (list). Data fetched Nov 17, 2026. Spot an error? Report it.