RunsOnMy ModelsGPUs & MacsBy VRAMAPI pricesVRAM calculatorNew

Can I run Llama 2 7b hf locally?

Llama 2 7b hf by NousResearch has 6.7B parameters. At 4-bit with an 8K context it needs about 8.3 GB of GPU memory. It fits on 65 of 65 GPUs and Macs we track. See all models that fit in 12 GB of VRAM.

8.3 GBQ4_K_M · 8K context
RTX 3080 10GBsmallest consumer GPU that fits
M1 16GBsmallest Mac that fits
4Kmax context
MHAattention 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/Llama-2-7b-hf-i1-GGUF where available (marked •), otherwise parameter count × bits per weight. How we calculate.

QuantQuality4K ctx
2-bit (Q2_K) •rough4.8 GB
3-bit (Q3_K_M) •usable5.5 GB
4-bit (Q4_K_M) •good, most popular6.3 GB
5-bit (Q5_K_M) •very good6.9 GB
6-bit (Q6_K) •near lossless7.6 GB
8-bit (Q8_0)effectively lossless9.1 GB
16-bit (F16/BF16)original weights15.0 GB

Which GPUs and Macs run Llama 2 7b hf?

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 GBFitsQ6_K9.6 GB65details
RTX 4060 Ti 16GB16 GBFitsQ8_011.1 GB40details
RTX 309024 GBFitsF1617.0 GB69details
RTX 409024 GBFitsF1617.0 GB75details
RTX 509032 GBFitsF1617.0 GB133details
RTX 5070 Ti16 GBFitsQ8_011.1 GB125details
RX 7900 XTX24 GBFitsF1617.0 GB71details
Mac M4 Pro 48GB48 GBFitsF1617.0 GB20details
Mac M4 Max 128GB128 GBFitsF1617.0 GB41details
Mac M3 Ultra 512GB512 GBFitsF1617.0 GB61details
DGX Spark128 GBFitsF1617.0 GB20details
RTX 407012 GBFitsQ6_K9.6 GB91details
RTX 5060 Ti 16GB16 GBFitsQ8_011.1 GB63details
RTX 40608 GBFitsQ2_K6.8 GB107details
Arc B58012 GBFitsQ6_K9.6 GB82details
RTX 3060 Ti8 GBFitsQ2_K6.8 GB177details
RTX 30708 GBFitsQ2_K6.8 GB177details
RTX 3080 10GB10 GBFitsQ4_K_M8.3 GB186details
RTX 3080 12GB12 GBFitsQ6_K9.6 GB165details
RTX 3090 Ti24 GBFitsF1617.0 GB75details
RTX 4060 Ti 8GB8 GBFitsQ2_K6.8 GB114details
RTX 4070 Super12 GBFitsQ6_K9.6 GB91details
RTX 4070 Ti12 GBFitsQ6_K9.6 GB91details
RTX 4070 Ti Super16 GBFitsQ8_011.1 GB94details
RTX 408016 GBFitsQ8_011.1 GB100details
RTX 4080 Super16 GBFitsQ8_011.1 GB103details
RTX 50608 GBFitsQ2_K6.8 GB177details
RTX 507012 GBFitsQ6_K9.6 GB122details
RTX 508016 GBFitsQ8_011.1 GB134details
2x RTX 30902×24 GBFitsF1617.0 GB69details
2x RTX 40902×24 GBFitsF1617.0 GB75details
2x RTX 50902×32 GBFitsF1617.0 GB133details
RTX A600048 GBFitsF1617.0 GB57details
RTX 6000 Ada48 GBFitsF1617.0 GB71details
RTX PRO 6000 Blackwell96 GBFitsF1617.0 GB133details
L40S48 GBFitsF1617.0 GB64details
A100 40GB40 GBFitsF1617.0 GB115details
A100 80GB80 GBFitsF1617.0 GB151details
H100 PCIe80 GBFitsF1617.0 GB148details
H100 SXM80 GBFitsF1617.0 GB249details
H200141 GBFitsF1617.0 GB356details
RX 76008 GBFitsQ2_K6.8 GB114details
RX 7700 XT12 GBFitsQ6_K9.6 GB78details
RX 7800 XT16 GBFitsQ8_011.1 GB87details
RX 7900 GRE16 GBFitsQ8_011.1 GB80details
RX 7900 XT20 GBFitsF1617.0 GB59details
RX 907016 GBFitsQ8_011.1 GB89details
RX 9070 XT16 GBFitsQ8_011.1 GB89details
Ryzen AI Max+ 395 128GB128 GBFitsF1617.0 GB19details
Arc A770 16GB16 GBFitsQ8_011.1 GB78details
Mac M1 16GB16 GBFitsQ6_K9.6 GB12details
Mac M2 24GB24 GBFitsQ8_011.1 GB14details
Mac M3 24GB24 GBFitsQ8_011.1 GB14details
Mac M4 16GB16 GBFitsQ6_K9.6 GB22details
Mac M4 32GB32 GBFitsF1617.0 GB8.9details
Mac M1 Pro 32GB32 GBFitsF1617.0 GB15details
Mac M3 Pro 36GB36 GBFitsF1617.0 GB11details
Mac M4 Pro 24GB24 GBFitsQ8_011.1 GB38details
Mac M4 Pro 64GB64 GBFitsF1617.0 GB20details
Mac M1 Max 64GB64 GBFitsF1617.0 GB30details
Mac M2 Max 96GB96 GBFitsF1617.0 GB30details
Mac M3 Max 128GB128 GBFitsF1617.0 GB30details
Mac M4 Max 64GB64 GBFitsF1617.0 GB41details
Mac M2 Ultra 192GB192 GBFitsF1617.0 GB59details
Mac M3 Ultra 256GB256 GBFitsF1617.0 GB61details

Questions

How much VRAM does Llama 2 7b hf need?

About 8.3 GB at 4-bit (Q4_K_M) with an 8K context: 3.8 GB of weights (real GGUF file size), 4.0 GB of KV cache and 0.5 GB of runtime overhead. At 8-bit it needs about 11.1 GB.

What is the smallest GPU that runs Llama 2 7b hf?

At Q4_K_M and 8K context, the smallest single consumer GPU in our table that fits it is the RTX 3080 10GB (10 GB).

Can I run Llama 2 7b hf 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.

Hermes 3 Llama 3.1 8B

8.0B · 6.2 GB at 4-bit

Meta Llama 3 8B Instruct

8.0B · 6.2 GB at 4-bit

Meta Llama 3.1 8B Instruct

8.0B · 6.2 GB at 4-bit

deepseek coder 6.7b instruct

6.7B · 8.3 GB at 4-bit

deepseek coder 6.7b base

6.7B · 8.3 GB at 4-bit

deepseek coder 7b instruct v1.5

6.9B · 8.3 GB at 4-bit

Model facts

Hugging Face repoNousResearch/Llama-2-7b-hf
Parameters6,738,417,664 (6.7B)
Layers / hidden size32 / 4096
AttentionMHA; 32 heads, 32 KV heads, head dim 128
KV cache per token (FP16)512.0 KiB
Max context4,096 tokens
License—
Released on HFJul 18, 2023
Downloads (30d) / likes90,189 / 178

Sources: Hugging Face model API · config.json (NousResearch/Llama-2-7b-hf) · GGUF files · GPU specs from vendor pages (list). Data fetched Nov 17, 2026. Spot an error? Report it.