DeepSeek-V2-Lite-Chat

DeepSeek checkpoint deepseek-ai/DeepSeek-V2-Lite-Chat at revision 85864749cd61. Architecture deepseek_v2, attention mla, calculator support unsupported.

Total parameters: 15,700,000,000 · active 2,400,000,000 (weights still use the total unless you model expert offload, which this version does not).

Layers 27, query heads 16, KV heads 16, head dim 128, recorded context 163,840 tokens.

Fetched 2026-09-19, verified 2026-09-19. Completeness: partial.

Memory by quantization and context

Concurrent sequences = 1. KV cache defaults to FP16. Range includes documented runtime extras for the listed engine. These HTML figures use the same core as the interactive calculator.

Quantization	2,048 tokens	8,192 tokens	32,768 tokens
BF16 safetensors	unsupported	unsupported	unsupported

Sources

DeepSeek-V2-Lite-Chat model card — MLA + MoE. 15.7B total / 2.4B active. Full KV formula is not applied.

Calculator

Pick a model and settings to see memory demand.

ModelHardwareQuantizationContext length (cached tokens)Concurrent sequences

Advanced settings

KV cache typeEngineDeploymentDevice countOverhead scenarioSystem / display reserve

GiB (2^30 bytes)

Budget already excludes system reserveExtra safety marginCustom bits per weight Include heavier prefill workspaceDisplay unit

Custom model structure

Import local config.json

Parsed in the browser. The file is not uploaded.

Public Hugging Face repoRevision

Only huggingface.co, config.json, no token. Private repos are rejected.

Total parametersActive parametersLayersQuery headsKV headsHead dimension

Estimate

Unsupported / incomplete

The architecture or split mode is out of scope. Weight figures may still be shown.

Estimated memory range: 29.26 GiB – 29.26 GiB

Weights use selected file metadata. Runtime extras are still estimated.

Weights29.26 GiB

KV cache (raw)0.00 GiB

KV extras (quant metadata / packing)0.00 GiB

Activation / workspace0.00 GiB

Engine reserve0.00 GiB

Safety margin0.00 GiB

Software compatibility

Not verified. Enough memory does not mean this engine, quant, and OS will run. Support does not mean it will be fast.

Architecture mla is not fully supported for KV estimates.
MLA KV is not estimated with the GQA formula.

Memory headroom does not prove this engine, quantization, and OS combination is supported.

Assumptions that affect this result

Full-resident weight estimates use total parameters, including inactive MoE experts.
This MoE lists 15700000000 total parameters and 2400000000 active parameters. Active-parameter compute is not used as the weight-residency figure.
Weight bytes come from the selected weight files only, not from every file in the repository.
Download file size is not peak GPU memory. Runtime layout, allocator padding, and KV are extra.
Counted files: model.safetensors.index.json metadata.total_size.
Context budget is the number of cached tokens per sequence (prompt plus reserved generation).
Concurrent sequences share one model replica; weights are not multiplied by sequence count.
KV dtype is independent of weight quantization unless you change it.

Calculation method