HF Model & Benchmark Matrix

24,078 variants · 14,984 scores · 120 benchmarks

Qwen3.5-9B

Qwen/Qwen3.5-9B

kind
original
quant
bf16 (16-bit)
family
qwen
modality
image-text-to-text
root model
Qwen/Qwen3.5-9B-Base
provider
Qwen
license
apache-2.0
language
—
context
262,144
library
transformers
vRAM (GB)
—
size on disk
—
downloads
11,043,117
likes
1,760
created
—
last modified
2026-03-02T00:51:43+00:00 (2026)
huggingface
https://huggingface.co/Qwen/Qwen3.5-9B
tags
transformerssafetensorsqwen3_5image-text-to-textconversationalbase_model:Qwen/Qwen3.5-9B-Basebase_model:finetune:Qwen/Qwen3.5-9B-Baselicense:apache-2.0eval-resultsendpoints_compatibledeploy:azuredeploy:sagemakerregion:us

Benchmark scores (0)

no scores extracted for this variant.

Discussions (63)

titleauthorstatuscommentstype
Update README.mdblaze-aura69open0disc
Listed on OpenModelMapduola15open0disc
Add MDPBench private evaluation resultDelores-Linopen0disc
Qwen3.5-9B brain atlas activation census, OV-circuits, and capability fence (32layersjuiceb0xc0deopen0disc
Multilingual powerhouse — testing for mobile deployment3morixdopen0disc
Listed on OpenModelMapduola15open0disc
Deployed on HexGrid Cloud: 1x RTX 5090 + Qwen3.5 9B BF16 — 1280 tok/s peak, then TTFT goes from 0.7s to 18s, ShareGPT, concurrency 16–128hexgridcloudopen0disc
Add EvalEval community eval resultsEvalEvalBotclosed0disc
MTP provides no throughput improvement on Apple Silicon Metal (llama.cpp b9330) — net loss at every configurationskaman5open0disc
Add MMMU-Pro evaluation resultSaylorTwiftopen0disc
Question about OSWorld-Verified evaluation settings for Qwen3.5-9Bgooncoopen0disc
Add ParseBench evaluation resultsboyang-runllamaopen0disc
Qwen3.5willowoodsdraft0disc
[Docs] Conflicting sampling parameters in README.mdtrgtszropen0disc
Add MDPBench evaluation resultsDelores-Linopen0disc
有没有直接用transformer库的用法示例?tangjiuqi097open0disc
fix chat template to avoid empty historical `<think>` blockslatent-variableopen0disc
Can we have a FP8 version?drguolaiopen0disc
Upload generation_config.jsonManojkumar96open0disc
finetuninghuu-ontocordopen0disc
Update README.mdIFthisisrealitynbdsopen0disc
Delete README.mdManiche23open0disc
Will ai completely replace humansokateunopen0disc
Create 123Alex-Neuralopen0disc
Qwen3.5 14Bprogram-19closed0disc
Proposal: new chat_template_arg `enable_history_reasoning` for reusing prompt cache among querys within Agents .Abioyopen0disc
Tool calling issuesvince62sclosed0disc
Add ScreenSpot-Pro evaluation resultmerveopen0disc
Add MathArena evaluation result for hmmt/hmmt_feb_2026JasperDekoninckopen0disc
Add MathArena evaluation result for aime/aime_2026JasperDekoninckopen0disc
Is the Qwen3VL inference guide applicable to Qwen3.5?summerishereopen0disc
Upload 35bbidulgi99closed0disc
Create generation_config.jsonjalolaopen0disc
nobodyMiggyjpopen0disc
mapemamapemaclosed0disc
anuoluwapomapemaopen0disc
for the love of action modelsirmyyopen0disc
Just askinglevuci88open0disc
greetingahmafc123open0disc
kamotebalatubs123open0disc
ilhmop09cryptosxionopen0disc
Update README.mdcryptosxionopen0disc
mr cleanghetraopen0disc
Running the model using HF backendsahilj2701open0disc
life as a journeybdfsparklesopen0disc
tryingyoungbyuuopen0disc
Update README.mdkerasakitopen0disc
possible to run on RTX 2060 8GB + 32GB RAM DDR4?krigetaopen0disc
No generation_config.json?deep1401open0disc
Now we waiting for Qwen4-10BNaster17open0disc
How to disable or reduce thinkingAyoubChLinopen0disc
Giberish response when using the model with transformers library with an imagejluixjuradoopen0disc
Can‘t run AutoModelForSequenceClassification for Qwen3.5-9BHanyuYuzuruopen0disc
Tool call stops in middle of the conversation, doesn't finish the complete taskTechCarbasaopen0disc
How to reproduce the results of Android World?llmmlmanopen0disc
No tool call return when calling qwen3.5 9bcppowboyclosed0disc
how does it perform on SWE-Bench or in general for agentic codingmayankiit04open0disc
Installation Video and Testing - Step by Stepfahdmirzacopen0disc
Can we get a 9B-FP8 version nextkqopen0disc
DGX SPARK VLLM RESULTSRGMC98open0disc
value error, The checkpoint you are trying to load has model type `qwen3_5`cesinsingaporeopen0disc
Add GPQA Diamond evaluation resultSaylorTwiftopen0disc
new soto for denseox-oxopen0disc
Hugging Face model & benchmark data scraped & matrix-ified. 24,078 variants · 14,984 scores · 120 benchmarks.