All models

DeepSeek / LANGUAGE MODEL

DeepSeek-Coder-V2-Lite-Base VRAM requirements

Explore memory tradeoffs with sourced model data and transparent estimates.

Q4_K_M estimate

11.1 GiB

at 8,192 tokens

FP16 estimate

33.9 GiB

at 8,192 tokens

Suggested GPU

1 × RTX 3060 12 GiB

Model architecture

Model computation diagram
  1. Token embedding · 2048
  2. 27 layers · MLA · MoE 6/64 experts active
  3. Output head → token scores
Parameters15,706,484,224 (15.71B)
Layers27
Attention heads16
KV heads16
Head dimension192
Hidden size2048
Maximum context163,840 tokens
NotesStandard transformer KV-cache estimate.

Minimum hardware and family comparison

Estimates use batch 1, an FP16 KV cache and 12% headroom. Multi-GPU capacity assumes even sharding and requires a compatible runtime.

DevicesRequired totalSmallest fitting device exampleUsable per device
111.13 GiBRTX 3060 12 GiB12.0 GiB
211.46 GiBGTX 1060 6 GiB6.0 GiB
412.14 GiBGTX 1050 Ti 4 GiB4.0 GiB
813.48 GiBGTX 1060 3 GiB3.0 GiB

Quantization comparison

Batch 1 · FP16 KV cache · 1 GPU · totals include safety margin. Contexts beyond the model maximum are hypothetical and marked *.

PrecisionWeightsTotal @ 4kTotal @ 8kTotal @ 32k
FP16 / BF1616.0005 bpw · FP16 GGUF reference29.26 GiB33.74 GiB33.87 GiB34.67 GiB
FP88 bpw · Nominal bit width14.63 GiB17.36 GiB17.49 GiB18.29 GiB
INT88 bpw · Nominal bit width14.63 GiB17.36 GiB17.49 GiB18.29 GiB
NVFP66 bpw · Nominal bit width10.97 GiB13.26 GiB13.39 GiB14.19 GiB

GPU architecture and runtime requirements

NVIDIA architectureAttention backendPublished environment or release
Ampere / AdaFlashAttention-2CUDA ≥ 12.0 · PyTorch ≥ 2.2 · FP16 / BF16
Hopper (H100 / H800)FlashAttention-3CUDA ≥ 12.3
Blackwell (B200)FlashAttention-4 · CuTeDSLPyTorch 2.7 introduced Blackwell support in its CUDA 12.8 builds

Transformers version is checkpoint-specific. Model card · Kernel compatibility · PyTorch build notes

Recommended GPUs by memory fit

Q4_K_M @ 8k context, batch 1, FP16 KV cache. Smallest nominal memory first; green results only. Check backend and precision support separately.

Sources & verification

Try your configuration

Your workload

Workload

Precision of the weights held in memory.

Longer contexts make cache precision more important.

Runtime settingsRoofline estimate

Your hardware

Input settings

Sequences per batch. Serving concurrency and throughput require separate measurements.

Advanced runtime settings
Advanced · Model architecture
Layers
27
Attention heads
16
KV heads
16
Head dimension
192
Hidden size
2,048
Maximum context (tokens)
163,840
View model configuration

Performance and memory results

46.4%Memory use

Memory estimate

11.1 GiB needed

Fits12.9 GiB to spare24.0 GiB available

The estimated workload fits with the safety margin included.

Estimated GPU rental · $0.44/h · Per-GPU quote × devices

Generation speed
~68 token/s
Time per token
~14.7 ms
Time to first token
~0.279 s
DeepSeek-Coder-V2-Lite-Base

RTX 4090 24 GiB

Q4_K_M · FP16 KV

Model specifications
Memory breakdownGiB

Vertical line = usable hardware memory

  • Weights (Q4_K_M)8.95 GiB
  • KV cache (FP16)0.24 GiB
  • Runtime overhead0.75 GiB
  • CUDA Graph capture0.00 GiB · 0.0%
  • Safety margin1.19 GiB
Framework overhead breakdown

Planning allocations. Actual driver, library and allocator use varies by runtime.

Device runtime and driver
0.50 GiB
Operator workspaces
0.20 GiB
CUDA Graph
0.00 GiB
Allocator base
0.05 GiB
Other runtime buffers and communication
0.00 GiB
Compare context lengths
Compare GPU capacity

Inference simulation

DeepSeek-Coder-V2-Lite-Base · Q4_K_M · RTX 4090 24 GiB

Prompt processing
0.28 s
Token generation
1.87 s
Estimated duration
2.15 s
Ready
Timing playback · sample text0 / 128 · 0 s

Press play to preview how the sample streams.

Measured latency percentiles and request throughput

Frequently asked questions

How much VRAM does DeepSeek-Coder-V2-Lite-Base need?

Our default Q4_K_M estimate at 8,192 context tokens, batch 1 and FP16 KV cache is 11.1 GiB including the safety margin. These are estimates, not measured guarantees.

Why does context length change memory use?

KV cache grows linearly with context length and batch size. We use KV heads for grouped-query attention and the recorded KV-cache layer count.

Does a green result guarantee compatibility?

No. Green means the estimated memory including safety headroom fits. Backend support, precision support, kernels and runtime allocation can differ. Confirm compatibility and measure your actual workflow.

Where do these estimates come from?

Architecture and file sizes are traced to model sources. The precision table identifies measured GGUF rates, block storage and nominal bit widths. The source data for DeepSeek-Coder-V2-Lite-Base was last verified on 2026-11-20. Memory estimates use GiB.