RunsOnMy ModelsGPUs & MacsAPI pricesVRAM calculatorNew

Can I run MiniMax H3 Image PE locally?

MiniMax H3 Image PE by MiniMaxAI has 27B parameters. At 4-bit with an 8K context it needs about 16.8 GB of GPU memory. It fits on 56 of 65 GPUs and Macs we track.

16.8 GBQ4_K_M · 8K context
RX 7900 XTsmallest consumer GPU that fits
M4 32GBsmallest Mac that fits
256Kmax 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 mradermacher/MiniMax-H3-Image-PE-GGUF where available (marked •), otherwise parameter count × bits per weight. How we calculate.

QuantQuality4K ctx8K ctx32K ctx128K ctx
2-bit (Q2_K) •rough11.1 GB11.3 GB12.8 GB18.8 GB
3-bit (Q3_K_M) •usable13.5 GB13.8 GB15.3 GB21.3 GB
4-bit (Q4_K_M) •good, most popular16.5 GB16.8 GB18.3 GB24.3 GB
5-bit (Q5_K_M) •very good19.0 GB19.3 GB20.8 GB26.8 GB
6-bit (Q6_K) •near lossless21.7 GB21.9 GB23.4 GB29.4 GB
8-bit (Q8_0) •effectively lossless27.8 GB28.0 GB29.5 GB35.5 GB
16-bit (F16/BF16)original weights52.1 GB52.3 GB53.8 GB59.8 GB

Which GPUs and Macs run MiniMax H3 Image PE?

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 GBTightQ2_K11.3 GB34details
RTX 4060 Ti 16GB16 GBFitsQ2_K11.3 GB27details
RTX 309024 GBFitsQ5_K_M19.3 GB49details
RTX 409024 GBFitsQ5_K_M19.3 GB52details
RTX 509032 GBFitsQ6_K21.9 GB81details
RTX 5070 Ti16 GBFitsQ2_K11.3 GB84details
RX 7900 XTX24 GBFitsQ5_K_M19.3 GB50details
Mac M4 Pro 48GB48 GBFitsQ8_028.0 GB9.5details
Mac M4 Max 128GB128 GBFitsF1652.3 GB10.0details
Mac M3 Ultra 512GB512 GBFitsF1652.3 GB15details
DGX Spark128 GBFitsF1652.3 GB5.0details
RTX 407012 GBTightQ2_K11.3 GB47details
RTX 5060 Ti 16GB16 GBFitsQ2_K11.3 GB42details
RTX 40608 GBNo———details
Arc B58012 GBTightQ2_K11.3 GB43details
RTX 3060 Ti8 GBNo———details
RTX 30708 GBNo———details
RTX 3080 10GB10 GBNo———details
RTX 3080 12GB12 GBTightQ2_K11.3 GB85details
RTX 3090 Ti24 GBFitsQ5_K_M19.3 GB52details
RTX 4060 Ti 8GB8 GBNo———details
RTX 4070 Super12 GBTightQ2_K11.3 GB47details
RTX 4070 Ti12 GBTightQ2_K11.3 GB47details
RTX 4070 Ti Super16 GBFitsQ2_K11.3 GB63details
RTX 408016 GBFitsQ2_K11.3 GB67details
RTX 4080 Super16 GBFitsQ2_K11.3 GB69details
RTX 50608 GBNo———details
RTX 507012 GBTightQ2_K11.3 GB63details
RTX 508016 GBFitsQ2_K11.3 GB90details
2x RTX 30902×24 GBFitsQ8_028.0 GB33details
2x RTX 40902×24 GBFitsQ8_028.0 GB35details
2x RTX 50902×32 GBFitsF1652.3 GB33details
RTX A600048 GBFitsQ8_028.0 GB27details
RTX 6000 Ada48 GBFitsQ8_028.0 GB34details
RTX PRO 6000 Blackwell96 GBFitsF1652.3 GB33details
L40S48 GBFitsQ8_028.0 GB30details
A100 40GB40 GBFitsQ8_028.0 GB54details
A100 80GB80 GBFitsF1652.3 GB37details
H100 PCIe80 GBFitsF1652.3 GB37details
H100 SXM80 GBFitsF1652.3 GB61details
H200141 GBFitsF1652.3 GB88details
RX 76008 GBNo———details
RX 7700 XT12 GBTightQ2_K11.3 GB40details
RX 7800 XT16 GBFitsQ2_K11.3 GB58details
RX 7900 GRE16 GBFitsQ2_K11.3 GB54details
RX 7900 XT20 GBFitsQ4_K_M16.8 GB48details
RX 907016 GBFitsQ2_K11.3 GB60details
RX 9070 XT16 GBFitsQ2_K11.3 GB60details
Ryzen AI Max+ 395 128GB128 GBFitsF1652.3 GB4.7details
Arc A770 16GB16 GBFitsQ2_K11.3 GB52details
Mac M1 16GB16 GBNo———details
Mac M2 24GB24 GBFitsQ3_K_M13.8 GB7.5details
Mac M3 24GB24 GBFitsQ3_K_M13.8 GB7.5details
Mac M4 16GB16 GBNo———details
Mac M4 32GB32 GBFitsQ5_K_M19.3 GB6.2details
Mac M1 Pro 32GB32 GBFitsQ5_K_M19.3 GB10details
Mac M3 Pro 36GB36 GBFitsQ5_K_M19.3 GB7.8details
Mac M4 Pro 24GB24 GBFitsQ3_K_M13.8 GB21details
Mac M4 Pro 64GB64 GBFitsQ8_028.0 GB9.5details
Mac M1 Max 64GB64 GBFitsQ8_028.0 GB14details
Mac M2 Max 96GB96 GBFitsF1652.3 GB7.3details
Mac M3 Max 128GB128 GBFitsF1652.3 GB7.3details
Mac M4 Max 64GB64 GBFitsQ8_028.0 GB19details
Mac M2 Ultra 192GB192 GBFitsF1652.3 GB15details
Mac M3 Ultra 256GB256 GBFitsF1652.3 GB15details

Questions

How much VRAM does MiniMax H3 Image PE need?

About 16.8 GB at 4-bit (Q4_K_M) with an 8K context: 15.4 GB of weights (real GGUF file size), 0.5 GB of KV cache and 0.9 GB of runtime overhead. At 8-bit it needs about 28.0 GB.

What is the smallest GPU that runs MiniMax H3 Image PE?

At Q4_K_M and 8K context, the smallest single consumer GPU in our table that fits it is the RX 7900 XT (20 GB).

Can I run MiniMax H3 Image PE on a Mac?

Yes. At Q4_K_M with 8K context it fits on a Mac M4 32GB or larger, using the default macOS GPU memory limit.

Model facts

Hugging Face repoMiniMaxAI/MiniMax-H3-Image-PE
Parameters27,356,728,560 (27B)
Layers / hidden size64 / 5120
AttentionGQA (hybrid: 48 linear/state layers); 24 heads, 4 KV heads, head dim 256
KV cache per token (FP16)64.0 KiB
Max context262,144 tokens
Licenseother
Released on HFNov 8, 2026
Downloads (30d) / likes2,730 / 120

Sources: Hugging Face model API · config.json (MiniMaxAI/MiniMax-H3-Image-PE) · GGUF files · GPU specs from vendor pages (list). Data fetched Nov 17, 2026. Spot an error? Report it.