Qwen

Qwen3.5 9B

Qwen3.5 9B is an open-weights 9B-parameter model by Qwen. It holds an Intelligence score of 55.3 on the local.ai leaderboard (2026-08 snapshot).

open weights
55.3
local.ai Intelligence
Parameters
9B
Benchmarks (local.ai snapshot)
τ²-bench
98%
GAIA
55%
GDPval
47%
HuggingFace metadata · updated 2026-03-02
repo
Qwen/Qwen3.5-9B
license
apache-2.0
downloads
14,206,979
likes
1,817

Metadata refreshed from the HuggingFace API (snapshot 2026-08-15T15:53:42.405Z). Context length and long-form descriptions are the next pipeline stage.

Does it fit your hardware?

Detect or select your GPU first. Green = weights + context headroom, yellow = tight, grey = does not fit.

Q4_K_M 5.2GQ6_K 8.1GQ8_0 9.9GFP16 18G
Quantization≈ File sizeNote
IQ2_XXS2.4 GBExtreme compression, largest quality loss
IQ3_XXS3.2 GBVery small
Q3_K_M4.1 GBSmall
NVFP44.5 GBNVIDIA 4-bit
Q4_K_S4.9 GBCommon
Q4_K_M5.2 GBMost common sweet spot
Q5_K_M6.8 GBBalanced
Q6_K8.1 GBNear-lossless
Q8_09.9 GBNear-lossless
FP1618 GBOriginal weights

Actual files on HuggingFace

FileSize
model.safetensors-00003-of-00004.safetensors5.00 GB
model.safetensors-00002-of-00004.safetensors4.97 GB
model.safetensors-00001-of-00004.safetensors4.91 GB
model.safetensors-00004-of-00004.safetensors3.10 GB
GGUF: unsloth/Qwen3.5-9B-GGUF · 1,180,204 downloads
FileSize
Qwen3.5-9B-BF16.gguf16.69 GBdownload
Qwen3.5-9B-IQ4_NL.gguf5.00 GBdownload
Qwen3.5-9B-IQ4_XS.gguf4.81 GBdownload
Qwen3.5-9B-Q3_K_M.gguf4.35 GBdownload
Qwen3.5-9B-Q3_K_S.gguf4.02 GBdownload
Qwen3.5-9B-Q4_0.gguf5.01 GBdownload
Qwen3.5-9B-Q4_1.gguf5.44 GBdownload
Qwen3.5-9B-Q4_K_M.gguf5.29 GBdownload
Qwen3.5-9B-Q4_K_S.gguf5.02 GBdownload
Qwen3.5-9B-Q5_K_M.gguf6.13 GBdownload
Qwen3.5-9B-Q5_K_S.gguf5.92 GBdownload
Qwen3.5-9B-Q6_K.gguf6.95 GBdownload
GGUF: DavidAU/Qwen3.5-9B-The-Defiant-Fable-Uncensored-Heretic-NEO-IMATRIX-MAX-MTP-GGUF · 639,475 downloads
FileSize
Qwen3.5-9B-The-Defiant-Fable-Uncnr-Heretic-NEO-MAX-IQ2_M.gguf4.60 GBdownload
Qwen3.5-9B-The-Defiant-Fable-Uncnr-Heretic-NEO-MAX-IQ3_M.gguf5.23 GBdownload
Qwen3.5-9B-The-Defiant-Fable-Uncnr-Heretic-NEO-MAX-IQ4_NL.gguf6.16 GBdownload
Qwen3.5-9B-The-Defiant-Fable-Uncnr-Heretic-NEO-MAX-IQ4_XS.gguf5.96 GBdownload
Qwen3.5-9B-The-Defiant-Fable-Uncnr-Heretic-NEO-MAX-MTP-IQ2_M.gguf4.60 GBdownload
Qwen3.5-9B-The-Defiant-Fable-Uncnr-Heretic-NEO-MAX-MTP-IQ3_M.gguf5.33 GBdownload
Qwen3.5-9B-The-Defiant-Fable-Uncnr-Heretic-NEO-MAX-MTP-IQ4_NL.gguf6.29 GBdownload
Qwen3.5-9B-The-Defiant-Fable-Uncnr-Heretic-NEO-MAX-MTP-IQ4_XS.gguf6.08 GBdownload
Qwen3.5-9B-The-Defiant-Fable-Uncnr-Heretic-NEO-MAX-MTP-Q4_K_M.gguf6.50 GBdownload
Qwen3.5-9B-The-Defiant-Fable-Uncnr-Heretic-NEO-MAX-MTP-Q4_K_S.gguf6.23 GBdownload
Qwen3.5-9B-The-Defiant-Fable-Uncnr-Heretic-NEO-MAX-MTP-Q5_K_M.gguf7.30 GBdownload
Qwen3.5-9B-The-Defiant-Fable-Uncnr-Heretic-NEO-MAX-MTP-Q5_K_S.gguf7.15 GBdownload

What you save self-hosting this model

Uses the model's own API price when it exists; otherwise the closest commercial equivalent.

Waiting for hardware detection…

RunLocal presents third-party leaderboard data (local.ai / Exo Labs, 2026-08) with its own fit and cost estimates. Verify pricing, licenses and file sizes at the source before deploying a model.