unsloth / llama family / image-text-to-text

llama-4-scout

meta/llama-4-scout

GGUF Provisional scores Updated 17 Jun 2025

Run locally Smallest indexed GGUF / needs Ollama

ollama run hf.co/meta/llama-4-scout/Llama-4-Scout-17B-16E-Instruct-UD-TQ1_0.gguf

Benchmarks

Where it scores.

Methodology →

General Provisional

61.3

Coding Provisional

57.2

Math Provisional

57.2

Reasoning Provisional

57.2

Normalized 0–100 across indexed models.

Runnable variants

Pick a size.

27 indexed

Precision	Format	Size	Params	File
Q4 / smallest	GGUF	29.3 GB	—	Llama-4-Scout-17B-16E-Instruct-UD-TQ1_0.gguf
IQ1_S	GGUF	32.5 GB	—	Llama-4-Scout-17B-16E-Instruct-UD-IQ1_S.gguf
IQ1_M	GGUF	35.0 GB	—	Llama-4-Scout-17B-16E-Instruct-UD-IQ1_M.gguf
IQ2_XXS	GGUF	37.4 GB	—	Llama-4-Scout-17B-16E-Instruct-UD-IQ2_XXS.gguf
IQ2_M	GGUF	39.1 GB	—	Llama-4-Scout-17B-16E-Instruct-UD-IQ2_M.gguf
Q2_K	GGUF	39.6 GB	—	Llama-4-Scout-17B-16E-Instruct-Q2_K.gguf
Q2_K_L	GGUF	39.8 GB	—	Llama-4-Scout-17B-16E-Instruct-Q2_K_L.gguf
Q2_K_XL	GGUF	42.4 GB	—	Llama-4-Scout-17B-16E-Instruct-UD-Q2_K_XL.gguf
IQ3_XXS	GGUF	45.7 GB	—	Llama-4-Scout-17B-16E-Instruct-UD-IQ3_XXS.gguf
Q3_K_S	GGUF	46.7 GB	—	Llama-4-Scout-17B-16E-Instruct-Q3_K_S.gguf
Q3_K_XL	GGUF	49.0 GB	—	Llama-4-Scout-17B-16E-Instruct-UD-Q3_K_XL.gguf
Q3_K_M	GGUF	51.8 GB	—	Q3_K_M/Llama-4-Scout-17B-16E-Instruct-Q3_K_M-00001-of-00002.gguf
IQ4_XS	GGUF	57.7 GB	—	IQ4_XS/Llama-4-Scout-17B-16E-Instruct-IQ4_XS-00001-of-00002.gguf
IQ4_NL	GGUF	60.9 GB	—	IQ4_NL/Llama-4-Scout-17B-16E-Instruct-IQ4_NL-00001-of-00002.gguf
Q4_0	GGUF	61.2 GB	—	Q4_0/Llama-4-Scout-17B-16E-Instruct-Q4_0-00001-of-00002.gguf
Q4_K_S	GGUF	61.5 GB	—	Q4_K_S/Llama-4-Scout-17B-16E-Instruct-Q4_K_S-00001-of-00002.gguf
Q4_K_XL	GGUF	62.0 GB	—	UD-Q4_K_XL/Llama-4-Scout-17B-16E-Instruct-UD-Q4_K_XL-00001-of-00002.gguf
Q4_K_M	GGUF	65.4 GB	—	Q4_K_M/Llama-4-Scout-17B-16E-Instruct-Q4_K_M-00001-of-00002.gguf
Q4_1	GGUF	67.6 GB	—	Q4_1/Llama-4-Scout-17B-16E-Instruct-Q4_1-00001-of-00002.gguf
Q5_K_S	GGUF	74.3 GB	—	Q5_K_S/Llama-4-Scout-17B-16E-Instruct-Q5_K_S-00001-of-00002.gguf
Q5_K_M	GGUF	76.5 GB	—	Q5_K_M/Llama-4-Scout-17B-16E-Instruct-Q5_K_M-00001-of-00002.gguf
Q5_K_XL	GGUF	79.1 GB	—	UD-Q5_K_XL/Llama-4-Scout-17B-16E-Instruct-UD-Q5_K_XL-00001-of-00002.gguf
Q6_K	GGUF	88.4 GB	—	Q6_K/Llama-4-Scout-17B-16E-Instruct-Q6_K-00001-of-00002.gguf
Q6_K_XL	GGUF	94.1 GB	—	UD-Q6_K_XL/Llama-4-Scout-17B-16E-Instruct-UD-Q6_K_XL-00001-of-00002.gguf
Q8_0	GGUF	114.5 GB	—	Q8_0/Llama-4-Scout-17B-16E-Instruct-Q8_0-00001-of-00003.gguf
Q8_K_XL	GGUF	128.2 GB	—	UD-Q8_K_XL/Llama-4-Scout-17B-16E-Instruct-UD-Q8_K_XL-00001-of-00003.gguf
BF16	GGUF	215.6 GB	—	BF16/Llama-4-Scout-17B-16E-Instruct-BF16-00001-of-00005.gguf

Sizes are file sizes, not peak runtime memory.

Raw benchmarks

The inputs.

2 records

Source	Task	Raw	Normalized
ZeroEval	GPQA	57.2	57.2
ZeroEval	MMMU	69.4	69.4

From the Hugging Face model card

About this model.

Read it on Hugging Face ↗