RunsOnMy ModelsGPUs & MacsBy VRAMAPI pricesVRAM calculatorNew

Can I run granite 4.0 h tiny locally?

granite 4.0 h tiny by ibm-granite has 6.9B parameters (1.5B active per token, 64 experts, 6 used per token). At 4-bit with an 8K context it needs about 4.6 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.

4.6 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-4.0-h-tiny-GGUF where available (marked •), otherwise parameter count × bits per weight. How we calculate.

QuantQuality4K ctx8K ctx32K ctx128K ctx
2-bit (Q2_K) •rough3.0 GB3.1 GB3.2 GB4.0 GB
3-bit (Q3_K_M) •usable3.7 GB3.8 GB4.0 GB4.7 GB
4-bit (Q4_K_M) •good, most popular4.6 GB4.6 GB4.8 GB5.5 GB
5-bit (Q5_K_M) •very good5.2 GB5.3 GB5.5 GB6.2 GB
6-bit (Q6_K) •near lossless5.9 GB6.0 GB6.2 GB6.9 GB
8-bit (Q8_0) •effectively lossless7.5 GB7.5 GB7.7 GB8.5 GB
16-bit (F16/BF16) •original weights13.6 GB13.6 GB13.8 GB14.5 GB

Which GPUs and Macs run granite 4.0 h tiny?

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_07.5 GB231details
RTX 4060 Ti 16GB16 GBFitsF1613.6 GB98details
RTX 309024 GBFitsF1613.6 GB319details
RTX 409024 GBFitsF1613.6 GB344details
RTX 509032 GBFitsF1613.6 GB611details
RTX 5070 Ti16 GBFitsF1613.6 GB305details
RX 7900 XTX24 GBFitsF1613.6 GB327details
Mac M4 Pro 48GB48 GBFitsF1613.6 GB93details
Mac M4 Max 128GB128 GBFitsF1613.6 GB186details
Mac M3 Ultra 512GB512 GBFitsF1613.6 GB279details
DGX Spark128 GBFitsF1613.6 GB93details
RTX 407012 GBFitsQ8_07.5 GB323details
RTX 5060 Ti 16GB16 GBFitsF1613.6 GB153details
RTX 40608 GBFitsQ6_K6.0 GB226details
Arc B58012 GBFitsQ8_07.5 GB292details
RTX 3060 Ti8 GBFitsQ6_K6.0 GB371details
RTX 30708 GBFitsQ6_K6.0 GB371details
RTX 3080 10GB10 GBFitsQ8_07.5 GB487details
RTX 3080 12GB12 GBFitsQ8_07.5 GB584details
RTX 3090 Ti24 GBFitsF1613.6 GB344details
RTX 4060 Ti 8GB8 GBFitsQ6_K6.0 GB239details
RTX 4070 Super12 GBFitsQ8_07.5 GB323details
RTX 4070 Ti12 GBFitsQ8_07.5 GB323details
RTX 4070 Ti Super16 GBFitsF1613.6 GB229details
RTX 408016 GBFitsF1613.6 GB244details
RTX 4080 Super16 GBFitsF1613.6 GB251details
RTX 50608 GBFitsQ6_K6.0 GB371details
RTX 507012 GBFitsQ8_07.5 GB431details
RTX 508016 GBFitsF1613.6 GB327details
2x RTX 30902×24 GBFitsF1613.6 GB319details
2x RTX 40902×24 GBFitsF1613.6 GB344details
2x RTX 50902×32 GBFitsF1613.6 GB611details
RTX A600048 GBFitsF1613.6 GB262details
RTX 6000 Ada48 GBFitsF1613.6 GB327details
RTX PRO 6000 Blackwell96 GBFitsF1613.6 GB611details
L40S48 GBFitsF1613.6 GB295details
A100 40GB40 GBFitsF1613.6 GB530details
A100 80GB80 GBFitsF1613.6 GB695details
H100 PCIe80 GBFitsF1613.6 GB682details
H100 SXM80 GBFitsF1613.6 GB1142details
H200141 GBFitsF1613.6 GB1636details
RX 76008 GBFitsQ6_K6.0 GB239details
RX 7700 XT12 GBFitsQ8_07.5 GB277details
RX 7800 XT16 GBFitsF1613.6 GB213details
RX 7900 GRE16 GBFitsF1613.6 GB196details
RX 7900 XT20 GBFitsF1613.6 GB273details
RX 907016 GBFitsF1613.6 GB218details
RX 9070 XT16 GBFitsF1613.6 GB218details
Ryzen AI Max+ 395 128GB128 GBFitsF1613.6 GB87details
Arc A770 16GB16 GBFitsF1613.6 GB191details
Mac M1 16GB16 GBFitsQ8_07.5 GB44details
Mac M2 24GB24 GBFitsF1613.6 GB34details
Mac M3 24GB24 GBFitsF1613.6 GB34details
Mac M4 16GB16 GBFitsQ8_07.5 GB77details
Mac M4 32GB32 GBFitsF1613.6 GB41details
Mac M1 Pro 32GB32 GBFitsF1613.6 GB68details
Mac M3 Pro 36GB36 GBFitsF1613.6 GB51details
Mac M4 Pro 24GB24 GBFitsF1613.6 GB93details
Mac M4 Pro 64GB64 GBFitsF1613.6 GB93details
Mac M1 Max 64GB64 GBFitsF1613.6 GB136details
Mac M2 Max 96GB96 GBFitsF1613.6 GB136details
Mac M3 Max 128GB128 GBFitsF1613.6 GB136details
Mac M4 Max 64GB64 GBFitsF1613.6 GB186details
Mac M2 Ultra 192GB192 GBFitsF1613.6 GB273details
Mac M3 Ultra 256GB256 GBFitsF1613.6 GB279details

Questions

How much VRAM does granite 4.0 h tiny need?

About 4.6 GB at 4-bit (Q4_K_M) with an 8K context: 3.9 GB of weights (real GGUF file size), 0.1 GB of KV cache and 0.6 GB of runtime overhead. At 8-bit it needs about 7.5 GB.

What is the smallest GPU that runs granite 4.0 h tiny?

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 4.0 h tiny 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 3.3 8b instruct

8.2B · 6.3 GB at 4-bit

OLMoE 1B 7B 0924

6.9B · 5.4 GB at 4-bit

deepseek coder 7b instruct v1.5

6.9B · 8.3 GB at 4-bit

deepseek coder 6.7b instruct

6.7B · 8.3 GB at 4-bit

Model facts

Hugging Face repoibm-granite/granite-4.0-h-tiny
Parameters6,939,037,248 (6.9B) · ~1.5B active (computed from config)
Layers / hidden size40 / 1536
AttentionGQA (hybrid: 36 linear/state layers); 12 heads, 4 KV heads, head dim 128
KV cache per token (FP16)8.0 KiB
Max context131,072 tokens
Licenseapache-2.0
Released on HFSep 16, 2025
Downloads (30d) / likes30,052 / 217

Sources: Hugging Face model API · config.json (ibm-granite/granite-4.0-h-tiny) · GGUF files · GPU specs from vendor pages (list). Data fetched Nov 17, 2026. Spot an error? Report it.