RunsOnMy ModelsGPUs & MacsAPI pricesVRAM calculatorNew

Can I run DeepSeek Coder V2 Lite Instruct locally?

DeepSeek Coder V2 Lite Instruct by deepseek-ai has 16B parameters (2.7B active per token, 64 experts, 6 used per token). At 4-bit with an 8K context it needs about 10.5 GB of GPU memory. It fits on 65 of 65 GPUs and Macs we track.

10.5 GBQ4_K_M · 8K context
RTX 4060 Ti 16GBsmallest consumer GPU that fits
M2 24GBsmallest Mac that fits
160Kmax context
MLAattention type

Memory needed by quant and context length

GB of GPU memory = weights + KV cache (FP16) + runtime overhead. Weights use real file sizes from lmstudio-community/DeepSeek-Coder-V2-Lite-Instruct-GGUF where available (marked •), otherwise parameter count × bits per weight. How we calculate.

QuantQuality4K ctx8K ctx32K ctx128K ctx
2-bit (Q2_K)rough6.2 GB6.3 GB7.0 GB9.9 GB
3-bit (Q3_K_M)usable7.9 GB8.0 GB8.7 GB11.5 GB
4-bit (Q4_K_M) •good, most popular10.4 GB10.5 GB11.2 GB14.0 GB
5-bit (Q5_K_M) •very good11.8 GB11.9 GB12.6 GB15.4 GB
6-bit (Q6_K) •near lossless13.8 GB13.9 GB14.6 GB17.5 GB
8-bit (Q8_0) •effectively lossless16.3 GB16.4 GB17.1 GB19.9 GB
16-bit (F16/BF16)original weights30.0 GB30.1 GB30.8 GB33.6 GB

Which GPUs and Macs run DeepSeek Coder V2 Lite 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 GBFitsQ3_K_M8.0 GB277details
RTX 4060 Ti 16GB16 GBFitsQ5_K_M11.9 GB143details
RTX 309024 GBFitsQ8_016.4 GB331details
RTX 409024 GBFitsQ8_016.4 GB356details
RTX 509032 GBFitsQ8_016.4 GB633details
RTX 5070 Ti16 GBFitsQ5_K_M11.9 GB446details
RX 7900 XTX24 GBFitsQ8_016.4 GB339details
Mac M4 Pro 48GB48 GBFitsQ8_016.4 GB96details
Mac M4 Max 128GB128 GBFitsF1630.1 GB103details
Mac M3 Ultra 512GB512 GBFitsF1630.1 GB154details
DGX Spark128 GBFitsF1630.1 GB51details
RTX 407012 GBFitsQ3_K_M8.0 GB388details
RTX 5060 Ti 16GB16 GBFitsQ5_K_M11.9 GB223details
RTX 40608 GBFitsQ2_K6.3 GB273details
Arc B58012 GBFitsQ3_K_M8.0 GB351details
RTX 3060 Ti8 GBFitsQ2_K6.3 GB449details
RTX 30708 GBFitsQ2_K6.3 GB449details
RTX 3080 10GB10 GBFitsQ3_K_M8.0 GB584details
RTX 3080 12GB12 GBFitsQ3_K_M8.0 GB701details
RTX 3090 Ti24 GBFitsQ8_016.4 GB356details
RTX 4060 Ti 8GB8 GBFitsQ2_K6.3 GB289details
RTX 4070 Super12 GBFitsQ3_K_M8.0 GB388details
RTX 4070 Ti12 GBFitsQ3_K_M8.0 GB388details
RTX 4070 Ti Super16 GBFitsQ5_K_M11.9 GB335details
RTX 408016 GBFitsQ5_K_M11.9 GB357details
RTX 4080 Super16 GBFitsQ5_K_M11.9 GB367details
RTX 50608 GBFitsQ2_K6.3 GB449details
RTX 507012 GBFitsQ3_K_M8.0 GB517details
RTX 508016 GBFitsQ5_K_M11.9 GB478details
2x RTX 30902×24 GBFitsF1630.1 GB176details
2x RTX 40902×24 GBFitsF1630.1 GB189details
2x RTX 50902×32 GBFitsF1630.1 GB337details
RTX A600048 GBFitsF1630.1 GB144details
RTX 6000 Ada48 GBFitsF1630.1 GB180details
RTX PRO 6000 Blackwell96 GBFitsF1630.1 GB337details
L40S48 GBFitsF1630.1 GB162details
A100 40GB40 GBFitsF1630.1 GB292details
A100 80GB80 GBFitsF1630.1 GB383details
H100 PCIe80 GBFitsF1630.1 GB376details
H100 SXM80 GBFitsF1630.1 GB629details
H200141 GBFitsF1630.1 GB902details
RX 76008 GBFitsQ2_K6.3 GB289details
RX 7700 XT12 GBFitsQ3_K_M8.0 GB332details
RX 7800 XT16 GBFitsQ5_K_M11.9 GB311details
RX 7900 GRE16 GBFitsQ5_K_M11.9 GB287details
RX 7900 XT20 GBFitsQ8_016.4 GB283details
RX 907016 GBFitsQ5_K_M11.9 GB319details
RX 9070 XT16 GBFitsQ5_K_M11.9 GB319details
Ryzen AI Max+ 395 128GB128 GBFitsF1630.1 GB48details
Arc A770 16GB16 GBFitsQ5_K_M11.9 GB279details
Mac M1 16GB16 GBFitsQ3_K_M8.0 GB52details
Mac M2 24GB24 GBFitsQ6_K13.9 GB42details
Mac M3 24GB24 GBFitsQ6_K13.9 GB42details
Mac M4 16GB16 GBFitsQ3_K_M8.0 GB92details
Mac M4 32GB32 GBFitsQ8_016.4 GB42details
Mac M1 Pro 32GB32 GBFitsQ8_016.4 GB71details
Mac M3 Pro 36GB36 GBFitsQ8_016.4 GB53details
Mac M4 Pro 24GB24 GBFitsQ6_K13.9 GB115details
Mac M4 Pro 64GB64 GBFitsF1630.1 GB51details
Mac M1 Max 64GB64 GBFitsF1630.1 GB75details
Mac M2 Max 96GB96 GBFitsF1630.1 GB75details
Mac M3 Max 128GB128 GBFitsF1630.1 GB75details
Mac M4 Max 64GB64 GBFitsF1630.1 GB103details
Mac M2 Ultra 192GB192 GBFitsF1630.1 GB150details
Mac M3 Ultra 256GB256 GBFitsF1630.1 GB154details

Questions

How much VRAM does DeepSeek Coder V2 Lite Instruct need?

About 10.5 GB at 4-bit (Q4_K_M) with an 8K context: 9.7 GB of weights (real GGUF file size), 0.2 GB of KV cache and 0.6 GB of runtime overhead. At 8-bit it needs about 16.4 GB.

What is the smallest GPU that runs DeepSeek Coder V2 Lite Instruct?

At Q4_K_M and 8K context, the smallest single consumer GPU in our table that fits it is the RTX 4060 Ti 16GB (16 GB).

Can I run DeepSeek Coder V2 Lite Instruct on a Mac?

Yes. At Q4_K_M with 8K context it fits on a Mac M2 24GB or larger, using the default macOS GPU memory limit.

Model facts

Hugging Face repodeepseek-ai/DeepSeek-Coder-V2-Lite-Instruct
Parameters15,706,484,224 (16B) · ~2.7B active (computed from config)
Layers / hidden size27 / 2048
AttentionMLA; 16 heads, 16 KV heads, head dim 128
KV cache per token (FP16)30.4 KiB
Max context163,840 tokens
Licenseother
Released on HFJun 14, 2024
Downloads (30d) / likes86,414 / 714

Sources: Hugging Face model API · config.json (deepseek-ai/DeepSeek-Coder-V2-Lite-Instruct) · GGUF files · GPU specs from vendor pages (list). Data fetched Nov 17, 2026. Spot an error? Report it.