How much VRAM does InternLM2 20B Chat need?

About 13.4 GB at Q4_K_M with 4K of context and batch 1, rising to roughly 19.2 GB at 32K. That figure is weights plus KV cache plus a 10% activation buffer, calculated from the model's architecture rather than measured on a card.

Will InternLM2 20B Chat run on a 16GB GPU?

Yes — at Q4_K_M and 4K context it needs about 13.4 GB, which leaves 2.6 GB spare on a RTX 4060 Ti 16G, or on any of the other 16 GB cards here (16 in all). That is the smallest size in this index that clears it comfortably; 50 of 78 do.

Which quantization of InternLM2 20B Chat should I use?

Q5_K_M has the lowest published quality loss (1.4%), and Q4_K_M is the level most people run. All 2 levels in the index are Q4_K_M, Q5_K_M.

InternLM2 20B Chat

20B

Shanghai AI Lab

⚠ Superseded · Prefer GPT-OSS 20B

This index would suggest GPT-OSS 20B instead today — longer context (32K → 128K tokens). The numbers below are still accurate; they are just for a model you probably should not start with. Compare the two →

Mid-size InternLM2 with excellent Chinese comprehension. Fits 24GB at Q4.

⬇ 2.0K HF downloads♥ 8 likesbartowski/internlm2_5-20b-chat-GGUF· stats from 11/7/2026

Consumer GPUPro GPU

33K

Max Context

Quant Variants

GGUF Q5_K_M

Best Quality

+1.4%

Perplexity rise (best build)

Calculate VRAM Hugging Face Compare

Quantization Variants

Per-quant VRAM, quality loss, and inference speed on RTX 4090

Measured = run on this site · Estimated = not run here: calculated, or a figure this site has not verified · Community = public reports

Format	Level	BPW	VRAM	PPL Loss	Speed	Source	Actions
GGUF	Q4_K_M	4.85	13.4 GB	2.9%	78 tok/s	Estimated	Calc HF
GGUF	Q5_K_M	5.68	15.6 GB	1.4%	68 tok/s	Estimated	Calc HF

Which file each size is for

A file size is not a memory figure: the VRAM column is the calculator's total at 4K context — weights plus KV cache plus a 10% activation buffer. Units differ on purpose: file sizes are decimal GB, as Hugging Face lists them; the VRAM column is GiB (1024³ bytes) with 2% added to the weights, so 7.12 GB of file counts as about 6.8 GiB there. A vision projector (mmproj), where the repo ships one, is a separate download and is in neither number.

GGUF Q4_K_MSized at the calculator's generic rate for this level; no file was checked.
GGUF Q5_K_MSized at the calculator's generic rate for this level; no file was checked.

Similar models

Compare with InternLM2 7B

InternLM2 7B Chat

Shanghai AI Lab

Consumer GPUMac / Apple Silicon

Q4_K_M5.5 GB+3.1%perplexity rise

Strong bilingual (EN/ZH) 7B from Shanghai AI Lab. Competitive with Qwen 7B.

32B

Qwen2.5 32B Instruct

Alibaba Qwen2.5

Consumer GPUPro GPU

Q4_K_M21.7 GB+2.7%perplexity rise

Near-GPT-4 reasoning on a 24GB VRAM card (Q4_K_S). Groundbreaking value.

24B

Mistral Small 24B Instruct

Mistral AI

Consumer GPUPro GPU

Q4_K_M15.9 GB+2.9%perplexity rise

Mistral's efficient 24B. Strong multilingual; fits on 24GB with Q4.

34B

Yi 1.5 34B Chat

01.AI Yi

Consumer GPUPro GPU

Q4_K_M22.8 GB+2.8%perplexity rise

01.AI's strong bilingual (EN/ZH) model. Competitive with Qwen 32B.

Running InternLM2 20B Chat locally

At Q4_K_M and 4K of context, InternLM2 20B Chat needs about 13.4 GB — 11.5 GB of weights, 0.75 GB of KV cache and a 1.2 GB activation buffer. The smallest cards in this index that clear that comfortably are 16 GB ones — 16 of them, from the RTX 4060 Ti 16G up — and 50 of the 78 cards here do. These are calculated figures, not measurements: the estimate stops counting a card as comfortable at 88% of its VRAM, which is roughly the room a desktop session needs.

What longer context costs

Going from 4K to 32K adds about 5.25 GB, taking the total to 19.2 GB. Weights do not move with context — only the KV cache does, and it grows linearly, so this is the number to watch when planning for long documents. The model's native window is 32K; holding all of it at Q4_K_M would need about 19 GB.

Which build to download

This index only tracks InternLM2 20B Chat in GGUF, across 2 levels (Q4_K_M, Q5_K_M). Q5_K_M carries the lowest published perplexity loss at 1.4%. The fastest level listed is Q4_K_M at about 78 tok/s on an RTX 4090, batch 1 — an estimate, not a run on this site's hardware. GGUF runs on llama.cpp and Ollama across NVIDIA, AMD and Apple silicon; AWQ and GPTQ target vLLM on CUDA and ROCm; EXL2 is ExLlamaV2 — CUDA only, and now archived, so local use rather than serving.

Common questions

How much VRAM does InternLM2 20B Chat need?: About 13.4 GB at Q4_K_M with 4K of context and batch 1, rising to roughly 19.2 GB at 32K. That figure is weights plus KV cache plus a 10% activation buffer, calculated from the model's architecture rather than measured on a card.
Will InternLM2 20B Chat run on a 16GB GPU?: Yes — at Q4_K_M and 4K context it needs about 13.4 GB, which leaves 2.6 GB spare on a RTX 4060 Ti 16G, or on any of the other 16 GB cards here (16 in all). That is the smallest size in this index that clears it comfortably; 50 of 78 do.
Which quantization of InternLM2 20B Chat should I use?: Q5_K_M has the lowest published quality loss (1.4%), and Q4_K_M is the level most people run. All 2 levels in the index are Q4_K_M, Q5_K_M.

Where this model fits

Sized at Q4_K_M with a 4K context window, smallest card first. Comfortable means the estimate uses at most 88% of the memory.

Runs comfortably on: RTX 4060 Ti 16G16GB · +2.6 Radeon RX 7600 XT16GB · +2.6 Radeon RX 9060 XT 16G16GB · +2.6 RTX 5060 Ti 16G16GB · +2.6 Radeon RX 6900 XT16GB · +2.6
Tight but possible: Mac M3 Pro 18G18GB
Just misses: RTX 3060 12G+1.4GB over Radeon RX 6700 XT+1.4GB over
Ships in: GGUF
Try it yourself: Size it in the calculator Generate the run command

How to actually run this

Deployment guides for this model and this class of hardware.

This page's figures change when the model or the runtime does.Last updated 2026-11-07 RSS → /feed.xml