Leaderboard / Alibaba (Qwen)

Qwen3.5 9B

Name: Qwen3.5 9B
Price: 0.028 USD
Author: Alibaba (Qwen)

Alibaba (Qwen) · Qwen3.5 · released 2026-03-02 · data checked 2026-11-01

Qwen3.5 9B is a 9B-parameter dense model from Alibaba (Qwen) (China), released 2026-03-02 under the Apache-2.0 license. It scores 63.9 on the Small LM Index, #1 of 26 small models, with 1 independently measured benchmark. It needs about 6.4 GB of memory at 4-bit with an 8K-token context and costs $0.028 per million tokens through the cheapest API provider.

Small LM Index

63.9

#1 of 26 scored small models

Likely 58.4–69.0 · from 3 of 6 core benchmarks

Parameters

Dense

Price per 1M tokens

$0.028

$0.025 in / $0.036 out · blended 3:1

Context window

262K

tokens

Benchmarks

Each bar shows this model's score. The tick marks the best small model on that benchmark. Sources: 5 reported by Alibaba (Qwen), 1 independently measured. How we label sources.

MMLU-Pro Lab-reported Knowledge82.5

GPQA Diamond Lab-reported Knowledge81.7

LiveCodeBench Lab-reported Coding · LiveCodeBench v665.6

Class best: Granite 4.2 8B 73.2

τ²-Bench Lab-reported Agentic79.1

IFBench Lab-reported Instruction following64.5

Class best: Granite 4.2 8B 79.3

MMMU-Pro Independent Vision · Measured by Artificial Analysis69.2

Not reported: estimated

What the index model expects from this model's other scores. Not a measurement.

Humanity's Last Exam≈ 13

AIME≈ 88

SWE-bench Verified≈ 52

Specifications

Lab: Alibaba (Qwen) (CN · China)
Architecture: Dense transformer
Total parameters: 9B
Active parameters: All (dense)
Size class: Small
Context window: 262K tokens
Input modalities: text, image, video
Reasoning: Hybrid (thinking on/off)
License: Apache-2.0
Openness: Open weights
Memory to run: ~6.4 GB at 4-bit, 8K context (more sizes)
Hugging Face: 6.4M downloads (30 days) · 2.3K likes
API providers: 6 on OpenRouter
API price: $0.025 in / $0.036 out per 1M (OpenRouter, 2026-11-03)

Weights on Hugging Face ↗Try on OpenRouter ↗

Compare with

Qwen3.5 9B vs Nemotron Nano 9B v2 (8.9B, index 52.5)
Qwen3.5 9B vs Apertus 1.5 8B (8.9B, index 20.1)
Qwen3.5 9B vs MiniCPM-SALA (9.5B, index 52.6)
Qwen3.5 9B vs DeepSeek-R1-0528-Qwen3-8B (8.2B, index 53.1)

Ranked in

#4 in Best LLMs that run on an 8 GB GPU or laptop
#5 in Best LLMs that run on a 16 GB GPU
#7 in Best cheap LLM APIs (open models)

Sources

Something wrong here? Report it.

Run it locally

Uses unsloth/Qwen3.5-9B-GGUF (Q4_K_M). Swap the tag for another size: BF16, IQ4_NL, IQ4_XS, Q3_K_M, Q3_K_S, Q4_0, Q4_1, Q4_K_M….

Ollama

ollama run hf.co/unsloth/Qwen3.5-9B-GGUF:Q4_K_M

llama.cpp

llama-server -hf unsloth/Qwen3.5-9B-GGUF:Q4_K_M

Then open http://localhost:8080 for a chat window.

MLX (Mac)

pip install -U mlx-lm
mlx_lm.chat --model mlx-community/Qwen3.5-9B-MLX-4bit

Text only; for images use mlx-vlm.

vLLM (GPU server)

vllm serve Qwen/Qwen3.5-9B

Full-precision weights, for a server GPU.

Open in LM Studio

Memory needed: weights, context cache and runtime buffers.

Quantization	4K context	32K context	256K context
16-bit F16	19 GB	20 GB	27 GB
8-bit Q8_0	10 GB	11 GB	19 GB
4-bit Q4_K_M	6.3 GB	7.2 GB	15 GB

Check it against your GPU, Mac or phone →