RunsOnMy ModelsGPUs & MacsBy VRAMAPI pricesVRAM calculatorNew

Can I run granite 3.3 8b instruct locally?

granite 3.3 8b instruct by ibm-granite has 8.2B parameters. At 4-bit with an 8K context it needs about 6.3 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.

6.3 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-3.3-8b-instruct-GGUF where available (marked •), otherwise parameter count × bits per weight. How we calculate.

QuantQuality4K ctx8K ctx32K ctx128K ctx
2-bit (Q2_K) •rough4.0 GB4.6 GB8.4 GB23.4 GB
3-bit (Q3_K_M) •usable4.8 GB5.5 GB9.2 GB24.2 GB
4-bit (Q4_K_M) •good, most popular5.7 GB6.3 GB10.1 GB25.1 GB
5-bit (Q5_K_M) •very good6.5 GB7.1 GB10.9 GB25.9 GB
6-bit (Q6_K) •near lossless7.4 GB8.0 GB11.7 GB26.7 GB
8-bit (Q8_0) •effectively lossless9.2 GB9.8 GB13.6 GB28.6 GB
16-bit (F16/BF16) •original weights16.3 GB17.0 GB20.7 GB35.7 GB

Which GPUs and Macs run granite 3.3 8b 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_09.8 GB41details
RTX 4060 Ti 16GB16 GBFitsQ8_09.8 GB33details
RTX 309024 GBFitsF1617.0 GB57details
RTX 409024 GBFitsF1617.0 GB62details
RTX 509032 GBFitsF1617.0 GB110details
RTX 5070 Ti16 GBFitsQ8_09.8 GB103details
RX 7900 XTX24 GBFitsF1617.0 GB59details
Mac M4 Pro 48GB48 GBFitsF1617.0 GB17details
Mac M4 Max 128GB128 GBFitsF1617.0 GB33details
Mac M3 Ultra 512GB512 GBFitsF1617.0 GB50details
DGX Spark128 GBFitsF1617.0 GB17details
RTX 407012 GBFitsQ8_09.8 GB58details
RTX 5060 Ti 16GB16 GBFitsQ8_09.8 GB52details
RTX 40608 GBFitsQ4_K_M6.3 GB55details
Arc B58012 GBFitsQ8_09.8 GB53details
RTX 3060 Ti8 GBFitsQ4_K_M6.3 GB91details
RTX 30708 GBFitsQ4_K_M6.3 GB91details
RTX 3080 10GB10 GBFitsQ6_K8.0 GB113details
RTX 3080 12GB12 GBFitsQ8_09.8 GB105details
RTX 3090 Ti24 GBFitsF1617.0 GB62details
RTX 4060 Ti 8GB8 GBFitsQ4_K_M6.3 GB58details
RTX 4070 Super12 GBFitsQ8_09.8 GB58details
RTX 4070 Ti12 GBFitsQ8_09.8 GB58details
RTX 4070 Ti Super16 GBFitsQ8_09.8 GB77details
RTX 408016 GBFitsQ8_09.8 GB83details
RTX 4080 Super16 GBFitsQ8_09.8 GB85details
RTX 50608 GBFitsQ4_K_M6.3 GB91details
RTX 507012 GBFitsQ8_09.8 GB77details
RTX 508016 GBFitsQ8_09.8 GB111details
2x RTX 30902×24 GBFitsF1617.0 GB57details
2x RTX 40902×24 GBFitsF1617.0 GB62details
2x RTX 50902×32 GBFitsF1617.0 GB110details
RTX A600048 GBFitsF1617.0 GB47details
RTX 6000 Ada48 GBFitsF1617.0 GB59details
RTX PRO 6000 Blackwell96 GBFitsF1617.0 GB110details
L40S48 GBFitsF1617.0 GB53details
A100 40GB40 GBFitsF1617.0 GB95details
A100 80GB80 GBFitsF1617.0 GB125details
H100 PCIe80 GBFitsF1617.0 GB122details
H100 SXM80 GBFitsF1617.0 GB205details
H200141 GBFitsF1617.0 GB294details
RX 76008 GBFitsQ4_K_M6.3 GB58details
RX 7700 XT12 GBFitsQ8_09.8 GB50details
RX 7800 XT16 GBFitsQ8_09.8 GB72details
RX 7900 GRE16 GBFitsQ8_09.8 GB66details
RX 7900 XT20 GBFitsF1617.0 GB49details
RX 907016 GBFitsQ8_09.8 GB74details
RX 9070 XT16 GBFitsQ8_09.8 GB74details
Ryzen AI Max+ 395 128GB128 GBFitsF1617.0 GB16details
Arc A770 16GB16 GBFitsQ8_09.8 GB64details
Mac M1 16GB16 GBFitsQ6_K8.0 GB10details
Mac M2 24GB24 GBFitsQ8_09.8 GB12details
Mac M3 24GB24 GBFitsQ8_09.8 GB12details
Mac M4 16GB16 GBFitsQ6_K8.0 GB18details
Mac M4 32GB32 GBFitsF1617.0 GB7.3details
Mac M1 Pro 32GB32 GBFitsF1617.0 GB12details
Mac M3 Pro 36GB36 GBFitsF1617.0 GB9.2details
Mac M4 Pro 24GB24 GBFitsQ8_09.8 GB31details
Mac M4 Pro 64GB64 GBFitsF1617.0 GB17details
Mac M1 Max 64GB64 GBFitsF1617.0 GB24details
Mac M2 Max 96GB96 GBFitsF1617.0 GB24details
Mac M3 Max 128GB128 GBFitsF1617.0 GB24details
Mac M4 Max 64GB64 GBFitsF1617.0 GB33details
Mac M2 Ultra 192GB192 GBFitsF1617.0 GB49details
Mac M3 Ultra 256GB256 GBFitsF1617.0 GB50details

Questions

How much VRAM does granite 3.3 8b instruct need?

About 6.3 GB at 4-bit (Q4_K_M) with an 8K context: 4.6 GB of weights (real GGUF file size), 1.3 GB of KV cache and 0.5 GB of runtime overhead. At 8-bit it needs about 9.8 GB.

What is the smallest GPU that runs granite 3.3 8b 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 8b 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 8b instruct

8.2B · 6.3 GB at 4-bit

granite 3.2 8b instruct

8.2B · 6.3 GB at 4-bit

granite 4.2 8b

8.8B · 6.8 GB at 4-bit

Qwen3 8B

8.2B · 6.5 GB at 4-bit

DeepSeek R1 0528 Qwen3 8B

8.2B · 6.5 GB at 4-bit

Apertus 8B Instruct 2509

8.1B · 6.4 GB at 4-bit

Model facts

Hugging Face repoibm-granite/granite-3.3-8b-instruct
Parameters8,170,864,640 (8.2B)
Layers / hidden size40 / 4096
AttentionGQA; 32 heads, 8 KV heads, head dim 128
KV cache per token (FP16)160.0 KiB
Max context131,072 tokens
Licenseapache-2.0
Released on HFApr 9, 2025
Downloads (30d) / likes101,293 / 161

Sources: Hugging Face model API · config.json (ibm-granite/granite-3.3-8b-instruct) · GGUF files · GPU specs from vendor pages (list). Data fetched Nov 17, 2026. Spot an error? Report it.