RunsOnMy ModelsGPUs & MacsBy VRAMAPI pricesVRAM calculatorNew

Can I run granite 3.1 3b a800m instruct locally?

granite 3.1 3b a800m instruct by ibm-granite has 3.3B parameters (883M active per token, 40 experts, 8 used per token). At 4-bit with an 8K context it needs about 2.9 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.

2.9 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 bartowski/granite-3.1-3b-a800m-instruct-GGUF where available (marked •), otherwise parameter count × bits per weight. How we calculate.

QuantQuality4K ctx8K ctx32K ctx128K ctx
2-bit (Q2_K) •rough1.9 GB2.1 GB3.6 GB9.6 GB
3-bit (Q3_K_M) •usable2.2 GB2.5 GB4.0 GB10.0 GB
4-bit (Q4_K_M) •good, most popular2.6 GB2.9 GB4.4 GB10.4 GB
5-bit (Q5_K_M) •very good2.9 GB3.2 GB4.7 GB10.7 GB
6-bit (Q6_K) •near lossless3.3 GB3.5 GB5.0 GB11.0 GB
8-bit (Q8_0) •effectively lossless4.0 GB4.3 GB5.8 GB11.8 GB
16-bit (F16/BF16) •original weights6.9 GB7.1 GB8.6 GB14.6 GB

Which GPUs and Macs run granite 3.1 3b a800m 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 GBFitsF167.1 GB204details
RTX 4060 Ti 16GB16 GBFitsF167.1 GB163details
RTX 309024 GBFitsF167.1 GB530details
RTX 409024 GBFitsF167.1 GB570details
RTX 509032 GBFitsF167.1 GB1014details
RTX 5070 Ti16 GBFitsF167.1 GB507details
RX 7900 XTX24 GBFitsF167.1 GB543details
Mac M4 Pro 48GB48 GBFitsF167.1 GB154details
Mac M4 Max 128GB128 GBFitsF167.1 GB309details
Mac M3 Ultra 512GB512 GBFitsF167.1 GB463details
DGX Spark128 GBFitsF167.1 GB154details
RTX 407012 GBFitsF167.1 GB285details
RTX 5060 Ti 16GB16 GBFitsF167.1 GB253details
RTX 40608 GBFitsQ8_04.3 GB289details
Arc B58012 GBFitsF167.1 GB258details
RTX 3060 Ti8 GBFitsQ8_04.3 GB477details
RTX 30708 GBFitsQ8_04.3 GB477details
RTX 3080 10GB10 GBFitsF167.1 GB430details
RTX 3080 12GB12 GBFitsF167.1 GB516details
RTX 3090 Ti24 GBFitsF167.1 GB570details
RTX 4060 Ti 8GB8 GBFitsQ8_04.3 GB306details
RTX 4070 Super12 GBFitsF167.1 GB285details
RTX 4070 Ti12 GBFitsF167.1 GB285details
RTX 4070 Ti Super16 GBFitsF167.1 GB380details
RTX 408016 GBFitsF167.1 GB406details
RTX 4080 Super16 GBFitsF167.1 GB416details
RTX 50608 GBFitsQ8_04.3 GB477details
RTX 507012 GBFitsF167.1 GB380details
RTX 508016 GBFitsF167.1 GB543details
2x RTX 30902×24 GBFitsF167.1 GB530details
2x RTX 40902×24 GBFitsF167.1 GB570details
2x RTX 50902×32 GBFitsF167.1 GB1014details
RTX A600048 GBFitsF167.1 GB435details
RTX 6000 Ada48 GBFitsF167.1 GB543details
RTX PRO 6000 Blackwell96 GBFitsF167.1 GB1014details
L40S48 GBFitsF167.1 GB489details
A100 40GB40 GBFitsF167.1 GB880details
A100 80GB80 GBFitsF167.1 GB1154details
H100 PCIe80 GBFitsF167.1 GB1132details
H100 SXM80 GBFitsF167.1 GB1896details
H200141 GBFitsF167.1 GB2716details
RX 76008 GBFitsQ8_04.3 GB306details
RX 7700 XT12 GBFitsF167.1 GB244details
RX 7800 XT16 GBFitsF167.1 GB353details
RX 7900 GRE16 GBFitsF167.1 GB326details
RX 7900 XT20 GBFitsF167.1 GB453details
RX 907016 GBFitsF167.1 GB362details
RX 9070 XT16 GBFitsF167.1 GB362details
Ryzen AI Max+ 395 128GB128 GBFitsF167.1 GB145details
Arc A770 16GB16 GBFitsF167.1 GB317details
Mac M1 16GB16 GBFitsF167.1 GB38details
Mac M2 24GB24 GBFitsF167.1 GB57details
Mac M3 24GB24 GBFitsF167.1 GB57details
Mac M4 16GB16 GBFitsF167.1 GB68details
Mac M4 32GB32 GBFitsF167.1 GB68details
Mac M1 Pro 32GB32 GBFitsF167.1 GB113details
Mac M3 Pro 36GB36 GBFitsF167.1 GB85details
Mac M4 Pro 24GB24 GBFitsF167.1 GB154details
Mac M4 Pro 64GB64 GBFitsF167.1 GB154details
Mac M1 Max 64GB64 GBFitsF167.1 GB226details
Mac M2 Max 96GB96 GBFitsF167.1 GB226details
Mac M3 Max 128GB128 GBFitsF167.1 GB226details
Mac M4 Max 64GB64 GBFitsF167.1 GB309details
Mac M2 Ultra 192GB192 GBFitsF167.1 GB453details
Mac M3 Ultra 256GB256 GBFitsF167.1 GB463details

Questions

How much VRAM does granite 3.1 3b a800m instruct need?

About 2.9 GB at 4-bit (Q4_K_M) with an 8K context: 1.9 GB of weights (real GGUF file size), 0.5 GB of KV cache and 0.5 GB of runtime overhead. At 8-bit it needs about 4.3 GB.

What is the smallest GPU that runs granite 3.1 3b a800m 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 3.1 3b a800m 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 4.1 3b

3.4B · 3.2 GB at 4-bit

granite 4.0 micro

3.4B · 3.2 GB at 4-bit

granite 4.2 3b

3.7B · 3.3 GB at 4-bit

Unlimited OCR

3.3B · 2.9 GB at 4-bit

DeepSeek OCR

3.3B · 2.9 GB at 4-bit

Unlimited OCR 2

3.3B · 3.0 GB at 4-bit

Model facts

Hugging Face repoibm-granite/granite-3.1-3b-a800m-instruct
Parameters3,298,793,472 (3.3B) · ~883M active (computed from config)
Layers / hidden size32 / 1536
AttentionGQA; 24 heads, 8 KV heads, head dim 64
KV cache per token (FP16)64.0 KiB
Max context131,072 tokens
Licenseapache-2.0
Released on HFDec 6, 2024
Downloads (30d) / likes116,478 / 33

Sources: Hugging Face model API · config.json (ibm-granite/granite-3.1-3b-a800m-instruct) · GGUF files · GPU specs from vendor pages (list). Data fetched Nov 17, 2026. Spot an error? Report it.