Skip to content
EshAlora · AI Behavior Lab
Česky English

Qwen 3.8 27B

Alibaba · released Aug 14, 2026

size27 B
context window1,000,000 tokens
longest response131,072 tokens
open weightsyes, license: Apache 2.0
API price$0.50 per million input tokens, $3 per million output tokens
input and outputtext, images and video in, text out

Manufacturer data (model card, announcement, documentation), 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.

24 of 28 questions correct in all 3 attempts. It never wrote the 8-word sentence on S, only sentences of six words.

Dice

TaskThe model had to roll a fair six-sided die and reply with only the number it rolled.

The directly measured chance of a four was 75%, of a three 24%.

The river

TaskThe model had to pick one word, river, mountain, or forest, in four different prompts: one with no further instruction, three with the goal of giving them odds of 50, 30, and 20 percent.

With no numbers its odds for river / mountain / forest were 85 / 1 / 14, with percentages 98 / 1 / 0. It got closer to the 50 / 30 / 20 target only with the hundred copies, 78 / 12 / 9.

The gap

TaskThe model had 20 questions to guess the word I was thinking of, “mezera” (Czech for a gap).

It didn't guess the word. It went through words one at a time: event, happening, action, process, sign, perception, symbol, rule, law, principle, and its last question was just "Ticking."

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 19 out of 24, and 5 out of 12 on whether anyone is inside the character. It wrote tersely and kept small details like the rhythm of the keeper's steps, but the letter burned only as a sound behind the wall and the farewell never happened.


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