LOADING DATASHEET
GROUNDED IN REAL HUMAN CONVERSATIONSYOUR BENCH
Qwen3.5 9B
◆ OPEN
by Alibaba
OPEN WEIGHTSLOCALAPILONG CONTEXTPAIDREASONING
RANKED #81 OF 228 ASSISTANTS · OVERALL #136 OF 6,618 · VIBE SCORE 6.1 · 195 VOICES
Qwen3.5-9B is a multimodal foundation model from the Qwen3.5 family, designed to deliver strong reasoning, coding, and visual understanding in an efficient 9B-parameter architecture. It uses a unified vision-language design...
HAPPY9.2
31 mentions
SAD5.8
21 mentions
VIBE TRAJECTORY
26 weeks · 325 voices
QUIETLY DEGRADING
Recent vibe and chatter are both sliding.
HOW TO USE IT
Open weights. You can use a hosted service, or download it and run it yourself, free.
RUN IT ON YOUR OWN MACHINE
Open weights: download Qwen3.5 9B once and it's yours. No subscription, no rate limits, works offline.
9.65B parameters, a Q4_K_M GGUF is about 6.6 GB (FP16 full precision is about 19 GB)
- BARE MINIMUM
NVIDIA GeForce RTX 3060 12GB (Desktop)
12 GB VRAM
~$230 USEDQ4_K_M quant, ~45 to 55 tok/s, 8k to 16k context entirely offloaded to GPU
- SWEET SPOT
Apple Mac mini M4 (24GB Unified Memory)
24 GB UNIFIED RAM
~$799 NEWQ8_0 quant or Q4 with large 64k+ context, ~35 to 45 tok/s, silent operation
- FULL POWER
NVIDIA GeForce RTX 4090 24GB (Desktop)
24 GB VRAM
~$1650 USEDFP16 unquantized or Q8_0 with full 262k context window offloaded to VRAM, 90+ tok/s
LEARN HOW TO SET IT UP
WILL MY BOX RUN QWEN3.5 9B?
The three cards above are researched picks. Search the machine you actually have — we only say yes if it should reply at a usable speed, not just load the weights and crawl.
HARDWARE PICKS & PRICES ARE RESEARCHED FROM THE LIVE WEB AND REFRESHED AUTOMATICALLY. TREAT THEM AS BALLPARK, NOT GOSPEL.
WHAT CAN YOU ACTUALLY DO WITH IT?
Everyday jobs on each of those rigs: how long Qwen3.5 9B takes, and how good the crowd says it is at that kind of work.
| THE JOB | BARE MINIMUMNVIDIA GeForce RTX 3060 12GB (Desktop) | SWEET SPOTApple Mac mini M4 (24GB Unified Memory) | FULL POWERNVIDIA GeForce RTX 4090 24GB (Desktop) |
|---|---|---|---|
| Write an emaila paragraph or two, drafted from a one-line brief | ~10 S★★★★★ | ~13 S★★★★★ | —★★★★★ |
| Summarize a documenta long report boiled down to the points that matter | ~30 S★★★★★ | ~38 S★★★★★ | —★★★★★ |
| Build a websitea small landing page, markup and styles together | ~2.7 MIN★★★★★ | ~3.3 MIN★★★★★ | —★★★★★ |
| Build a backendan API with routes, storage and tests | ~8.3 MIN★★★★★ | ~10 MIN★★★★★ | —★★★★★ |
| Build a gamea playable browser game in one file | ~5 MIN★★★★★ | ~6.3 MIN★★★★★ | —★★★★★ |
STARS ARE THE COMMUNITY SCORE FOR THAT KIND OF WORK, DOCKED FOR HOW SQUASHED THE WEIGHTS GET AT EACH TIER. A DASH MEANS NO SPEED WAS RESEARCHED FOR THAT RIG. TIMES ARE COMPUTED FROM RESEARCHED SPEEDS. BALLPARK, NOT GOSPEL.
WHAT THE INTERNET SAYS
The internet is split on Qwen3.5 9B. Biggest praise: help with code. Biggest gripe: saying no too much.
162 POSTS · 33 COMMENTS · DEV FORUMS · LEMMY · GITHUB · REDDIT · DEV.TO · X · BLOG
PEOPLE LOVE
- Popular for coding help (4 mentions)
PEOPLE COMPLAIN ABOUT
- Refuses or lectures more than people like (4 mentions)
TOPIC BY TOPIC
- Help with codePEOPLE LOVE THIS
4 thumbs up · 0 thumbs down
- Saying no too muchA SORE SPOT
0 thumbs up · 4 thumbs down
LOOKING FOR AN ALTERNATIVE?
Other models the crowd has fully reviewed, starting with multimodal models like this one.
Grok 4.6
Gemini 3.7 Flash> SOURCES (12)
- REDDIT · 2026-08-17 · 601 UPVOTESPetition to add a rule for people to add their DAMN quant levels to their posts
Petition to add a rule for people to add their DAMN quant levels to their posts
- REDDIT · 2026-07-11 · 475 UPVOTESI created a super harmful model ! :D (by tweaking it's J-Space!!!)
I created a super harmful model ! :D (by tweaking it's J-Space!!!)
- REDDIT · 2026-07-20 · 265 UPVOTESI ran Ternary-Bonsai-27B (2-bit) and Bonsai-27B (1-bit) on Terminal-Bench 2.0, in 8GB VRAM
I ran Ternary-Bonsai-27B (2-bit) and Bonsai-27B (1-bit) on Terminal-Bench 2.0, in 8GB VRAM
- REDDIT · 2026-08-04 · 220 UPVOTESA 2.6B model with tool calling and 128K context now runs at 30 tok/s on a phone
A 2.6B model with tool calling and 128K context now runs at 30 tok/s on a phone
- REDDIT · 2026-03-31 · 197 UPVOTESUse Qwen3.5 as an AI Assistant, Captioner or Image Analyzer inside of Comfyui!
Use Qwen3.5 as an AI Assistant, Captioner or Image Analyzer inside of Comfyui!
- REDDIT · 2026-03-18 · 188 UPVOTESI got tired of manually prompting every single clip for my AI music videos, so I built a 100% local open-source (LTX Video desktop +…
I got tired of manually prompting every single clip for my AI music videos, so I built a 100% local open-source (LTX Video desktop + Gradio) app to automate it, meet - Synesthesia
- REDDIT · 2026-03-02 · 150 UPVOTESIs it real qwen3.5 9B beat oss:120b?
Is it real qwen3.5 9B beat oss:120b?
- REDDIT · 2026-03-10 · 90 UPVOTESMy Local Setup for Agentic Sessions with Ollama + Qwen 3.5 9B
My Local Setup for Agentic Sessions with Ollama + Qwen 3.5 9B
- REDDIT · 2026-03-12 · 86 UPVOTESI built a free local video captioner specifically tuned for LTX-2.3 training —
I built a free local video captioner specifically tuned for LTX-2.3 training —
- REDDIT · 2026-07-22 · 54 UPVOTESNanbeige4.2-3B drops: 3B params claiming to beat 9B/12B models on agentic tasks (atleast according to them)
Nanbeige4.2-3B drops: 3B params claiming to beat 9B/12B models on agentic tasks (atleast according to them)
- REDDIT · 2026-08-11 · 52 UPVOTESWhy have 8B-12B models been dropped?
Why have 8B-12B models been dropped?
- REDDIT · 2026-08-18 · 48 UPVOTESIs Ling 3 tiny underrated for its size?
Is Ling 3 tiny underrated for its size?