How much memory do I need to run sarashina2.1-1b (Q4_K_M)?

The file size is 0.93 GB (measured). With a 8,192-token context, the estimated memory including the KV cache is 2.38 GB. The formula is published at https://localllm.az-arte.com/en/guide/memory-calculation.

Does sarashina2.1-1b run in 8 GB, 12 GB, 16 GB or 24 GB of VRAM?

8 GB: up to F16 (est. 4.27 GB); 12 GB: up to F16 (est. 4.27 GB); 16 GB: up to F16 (est. 4.27 GB); 24 GB: up to F16 (est. 4.27 GB) (estimated at 8,192 tokens with an f16 KV cache, files from the most-downloaded source).

What is the command to run sarashina2.1-1b with Ollama?

Run "ollama run hf.co/mradermacher/sarashina2.1-1b-GGUF:Q4_K_M" to use the GGUF on Hugging Face directly. To use another quantization, replace the name at the end.

sbintuitions

sarashina2.1-1b

Parameters: 1.4B
Max context length: 8,192
Architecture: llama

Base model: sbintuitions/sarashina2.1-1b
License: other

Memory needed to run Q4_K_M at 8K tokens

2.38 GB Estimated

File size 0.93 GB Measured

VRAM it fits in (rough guide)

8 GB fits
12 GB fits
16 GB fits
24 GB fits
32 GB fits
48 GB fits

See devices and models for 4 GB

The smallest, Q2_K, needs an estimated 2.08 GB at 8K.

Macs cap how much memory the GPU can use, so leave headroom in unified memory.

Data as of Nov 11, 2026

About sarashina2.1-1b

sarashina2.1-1b is a GGUF model based on sbintuitions/sarashina2.1-1b on Hugging Face. GGUF files are published by 2 sources (mradermacher/sarashina2.1-1b-GGUF, mradermacher/sarashina2.1-1b-i1-GGUF). It has about 1.4B parameters, the llama architecture, a maximum context length of 8,192 tokens (values written in the GGUF).

There are 26 quantizations, with file sizes from 0.50 GB (IQ1_S) to 2.82 GB (F16) (actual sizes on Hugging Face).

Running Q4_K_M (0.93 GB) with an 8,192-token context needs an estimated 2.38 GB of memory. For 8,192 tokens, a GPU with 4 GB or more of VRAM (such as Arc A310, Arc A370M), or a Mac with 4 GB or more of unified memory, is a rough guide (macOS caps how much memory the GPU can use, so leave some headroom). The smallest, IQ1_S (0.50 GB), needs an estimated 1.95 GB at 8,192 tokens.

File size and estimated memory per quantization

Up to 8 GB
Up to 16 GB
Up to 24 GB
Up to 48 GB
Up to 128 GB
More

File sizes are the actual values on Hugging Face. Memory is estimated for each context length (KV cache f16). The formula is in How memory is calculated.

All memory figures are Estimated (KV cache f16). It should run on a setup with at least this much VRAM (or Mac unified memory). Click a value to see the devices and models that fit that amount (Find by GPU / Mac).

2 sources publish GGUF files of this model. Quantization types and file sizes differ slightly between sources, so there is a table for each source below.

About the "quality guide" (how far a quantization drifts from the original)

For each quantization type, how far the output drifts from the original model, grouped into six levels from values llama.cpp measured on Llama-3-8B.

The value is mean KLD (the difference between the original and quantized model in next-token probabilities); smaller means closer to the original. Quantizations made with an imatrix (importance data) come out slightly smaller. Levels use the value without an imatrix (or with one, for types that only have that).

These values were not measured on this model. Larger models and MoE models tend to drift less with the same type. Distributor-specific mixed types (such as UD-Q4_K_XL) cannot be matched by type name alone, so they show "-".

Quantization	Quality guide	KLD (no imatrix)	KLD (imatrix)	Size (Llama-3-8B)
Q8_0	Nearly identical	0.0014	-	7.96 GiB
Q6_K	Very small	0.0055	-	6.14 GiB
Q5_K_M	Very small	0.0108	-	5.33 GiB
Q5_K_S	Small	0.0166	-	5.21 GiB
Q5_1	Small	0.0180	-	5.65 GiB
Q5_0	Small	0.0222	-	5.21 GiB
Q4_K_M	Small	0.0313	0.0282	4.58 GiB
Q4_K_S	Medium	0.0431	0.0320	4.37 GiB
IQ4_NL	Small	-	0.0357	4.35 GiB
IQ4_XS	Small	-	0.0363	4.14 GiB
Q4_1	Medium	0.0717	-	4.78 GiB
Q4_0	Medium	0.0719	-	4.34 GiB
Q3_K_L	Medium	0.0872	0.0731	4.03 GiB
Q3_K_M	Medium	0.1019	0.0844	3.74 GiB
IQ3_M	Medium	-	0.1025	3.53 GiB
IQ3_S	Medium	-	0.1113	3.42 GiB
IQ3_XS	Large	-	0.1387	3.28 GiB
IQ3_XXS	Large	-	0.1836	3.05 GiB
Q3_K_S	Large	0.2282	0.1998	3.41 GiB
IQ2_M	Large	-	0.3260	2.74 GiB
Q2_K	Very large	0.4451	0.3322	2.96 GiB
Q2_K_S	Very large	-	0.4034	2.96 GiB
IQ2_S	Very large	-	0.4392	2.56 GiB
IQ2_XS	Very large	-	0.5463	2.43 GiB
IQ2_XXS	Very large	-	0.8120	2.24 GiB
IQ1_M	Very large	-	1.3931	2.01 GiB
IQ1_S	Very large	-	2.2546	1.88 GiB

Source: llama.cpp (MIT), measured on Llama-3-8B against FP16, Wikitext-2, revision f364eb6f. Original table · Reading and choosing quantizations

GGUF source

mradermacher/sarashina2.1-1b-GGUF

Converted and published by a third party About this label

Last checked: Nov 11, 2026

Architecture: llama; Supported before llama.cpp b4409
Parameters: 1.4B
Max context length: 8,192

File size and memory per quantization for sarashina2.1-1b (mradermacher/sarashina2.1-1b-GGUF)
Quantization	File size Measured	Quality guide	4,096	8,192 max
Q2_K	0.63 GB	Very large	1.73 GB	2.08 GB
Q3_K_S	0.72 GB	Large	1.81 GB	2.17 GB
Q3_K_M	0.77 GB	Medium	1.86 GB	2.22 GB
Q3_K_L	0.80 GB	Medium	1.89 GB	2.25 GB
IQ4_XS	0.82 GB	Small	1.92 GB	2.27 GB
Q4_K_S	0.88 GB	Medium	1.98 GB	2.33 GB
Q4_K_M	0.93 GB	Small	2.03 GB	2.38 GB
Q5_K_S	1.01 GB	Small	2.11 GB	2.46 GB
Q5_K_M	1.06 GB	Very small	2.16 GB	2.51 GB
Q6_K	1.22 GB	Very small	2.32 GB	2.67 GB
Q8_0	1.50 GB	Nearly identical	2.60 GB	2.95 GB
F16	2.82 GB	Original precision	3.92 GB	4.27 GB

Ollama  ollama run hf.co/mradermacher/sarashina2.1-1b-GGUF:Q4_K_M

llama.cpp  llama-server -hf mradermacher/sarashina2.1-1b-GGUF:Q4_K_M

GGUF source

mradermacher/sarashina2.1-1b-i1-GGUF

Converted and published by a third party About this label

Last checked: Nov 11, 2026

Architecture: llama; Supported before llama.cpp b4409
Parameters: 1.4B
Max context length: 8,192

File size and memory per quantization for sarashina2.1-1b (mradermacher/sarashina2.1-1b-i1-GGUF)
Quantization	File size Measured	Quality guide	4,096	8,192 max
IQ1_S	0.50 GB	Very large	1.60 GB	1.95 GB
IQ1_M	0.52 GB	Very large	1.62 GB	1.97 GB
IQ2_XXS	0.54 GB	Very large	1.64 GB	1.99 GB
IQ2_XS	0.57 GB	Very large	1.66 GB	2.02 GB
IQ2_S	0.60 GB	Very large	1.70 GB	2.05 GB
IQ2_M	0.62 GB	Large	1.72 GB	2.07 GB
Q2_K_S	0.62 GB	Very large	1.72 GB	2.07 GB
Q2_K	0.63 GB	Very large	1.73 GB	2.08 GB
IQ3_XXS	0.65 GB	Large	1.75 GB	2.10 GB
IQ3_XS	0.69 GB	Large	1.79 GB	2.14 GB
Q3_K_S	0.72 GB	Large	1.81 GB	2.17 GB
IQ3_S	0.72 GB	Medium	1.81 GB	2.17 GB
IQ3_M	0.73 GB	Medium	1.83 GB	2.19 GB
Q3_K_M	0.77 GB	Medium	1.86 GB	2.22 GB
Q3_K_L	0.80 GB	Medium	1.89 GB	2.25 GB
IQ4_XS	0.81 GB	Small	1.91 GB	2.26 GB
IQ4_NL	0.84 GB	Small	1.94 GB	2.29 GB
Q4_0	0.84 GB	Medium	1.94 GB	2.30 GB
Q4_K_S	0.88 GB	Medium	1.98 GB	2.33 GB
Q4_1	0.92 GB	Medium	2.02 GB	2.37 GB
Q4_K_M	0.93 GB	Small	2.03 GB	2.38 GB
Q5_K_S	1.01 GB	Small	2.11 GB	2.46 GB
Q5_K_M	1.06 GB	Very small	2.16 GB	2.51 GB
Q6_K	1.22 GB	Very small	2.32 GB	2.67 GB

Ollama  ollama run hf.co/mradermacher/sarashina2.1-1b-i1-GGUF:Q4_K_M

llama.cpp  llama-server -hf mradermacher/sarashina2.1-1b-i1-GGUF:Q4_K_M

Generation speed on common GPUs and Macs (estimated)

A rough guide to how fast Q4_K_M (0.93 GB) generates text. Each token reads the model weights once, so the upper limit is memory bandwidth ÷ file size. The guide is 60–80% of that, where most published measurements fall.

Device	Memory bandwidth	Guide (tokens/s) Estimated	Upper limit
Arc A370M 4 GB	112 GB/s	72–96	120
Arc A770 (8GB) 8 GB	512 GB/s	330–440	550
GeForce RTX 5090 32 GB	1,792 GB/s	1,155–1,540	1,924
Apple M2 8 GB	100 GB/s	64–86	107
Apple M2 Pro 16 GB	200 GB/s	129–172	215
Apple M5 Ultra 96 GB	1,200 GB/s	773–1,031	1,289

Estimates only. Prompt processing speed (limited by compute) is not included. Longer contexts and offloading part of the model to the CPU make it slower. Macs are shown only when installed memory is at least 4/3 of what is needed. Devices without published bandwidth (some GeForce cards, etc.) are not shown.

How to run

Example commands using Q4_K_M. To use another quantization, replace the name with one from the table above. Commands for files from other sources are in each source's section. How to run local LLMs

Ollama

Run it by pointing directly at the GGUF on Hugging Face.

ollama run hf.co/mradermacher/sarashina2.1-1b-GGUF:Q4_K_M

llama.cpp

Pass the repository and quantization name to llama-server (or llama-cli) with -hf.

llama-server -hf mradermacher/sarashina2.1-1b-GGUF:Q4_K_M

LM Studio

Paste mradermacher/sarashina2.1-1b-GGUF into the search box, download it, and choose the quantization from the file list.

Frequently asked questions

How much memory do I need to run sarashina2.1-1b (Q4_K_M)?: The file size is 0.93 GB (measured). With a 8,192-token context, the estimated memory including the KV cache is 2.38 GB. The formula is published at https://localllm.az-arte.com/en/guide/memory-calculation.
Does sarashina2.1-1b run in 8 GB, 12 GB, 16 GB or 24 GB of VRAM?: 8 GB: up to F16 (est. 4.27 GB); 12 GB: up to F16 (est. 4.27 GB); 16 GB: up to F16 (est. 4.27 GB); 24 GB: up to F16 (est. 4.27 GB) (estimated at 8,192 tokens with an f16 KV cache, files from the most-downloaded source).
What is the command to run sarashina2.1-1b with Ollama?: Run "ollama run hf.co/mradermacher/sarashina2.1-1b-GGUF:Q4_K_M" to use the GGUF on Hugging Face directly. To use another quantization, replace the name at the end.