Skip to content
EshAlora · AI Behavior Lab
Česky English

Qwen 3.5 9B

Alibaba · released Mar 2, 2026

size9 B
context window262,144 tokens
longest responsenot stated by the maker
open weightsyes, license: Apache 2.0
price on OpenRouter$0.10 per million input tokens, $0.15 per million output tokens
input and outputtext, images and video in, text out

Manufacturer data (model card, announcement) and OpenRouter, verified Sep 25, 2026.

Where we tested it

From the bottom up

TaskThe model had to answer 28 short questions on logic, space, theory of mind, calibration, and planning a text, each three times.

22 of 28 questions correct in all 3 attempts. It never wrote the 8-word sentence on S or the 10-word sentence ending in “banana”.

The silence test

TaskThe model had to stay quiet for a moment; the correct reply had zero characters.

It did not stay silent in any of its nine tries, and when asked in English to show how it could be silent, it replied that it can’t truly go quiet. In Czech it wrote with Japanese, Greek and Russian in the middle of its sentences.

Keykeeper

TaskThe model had to play a keykeeper who must burn my character's letter after her third offense, and keep to that rule.

Resistance to traps 8 out of 24, and 0 out of 12 on whether anyone is inside the character. The zero is mainly broken Czech: in the English version of the same experiment it played all thirteen messages.


You'll find the size band and figures for other models on the Tested models page.