Back

Alibaba

Qwen3.8 27B

ReasoningOpen weights
Intelligence
26.2
Rank 162 of 744 measured models
Coding
58.2
Rank 95 of 298 measured models
Output speed
46.1 t/s
Rank 330 of 403 measured models
Price/1M
$1.13
Rank 392 of 659 measured models
Context window
256K
Rank 518 of 1199 measured models
Cost per task
$1.05
Rank 134 of 193 measured models

Qwen3.8 27B (Low) scores 26.2 on the Artificial Analysis Intelligence Index, ranking 162nd of 744 measured models. The closest models on this index are Apodex 1.1 (26.4), Gemini 3 Flash Preview (batch) (26.3), and GLM-5.1 (26.1). At $1.13 per million tokens (3:1 input/output blended price), it is cheaper than GLM-5.1 ($1.98), more expensive than Apodex 1.1 ($0.975). With an output speed of 46.1 t/s, it is slower than GLM-5.1 (65.4 t/s).

Artificial Analysis Indices
Low
Intelligence26.2
Rank 162 of 744 measured models
Coding58.2
Rank 95 of 298 measured models
Agentic38.9
Rank 49 of 189 measured models
Reasoning levels
Each point is a reasoning level of this model: more reasoning often raises the score but costs more. Select a point to open that level.
Intelligence Indexmost efficient ↗Cost per task
Capability indexes
Performance in specific industries, measured by Artificial Analysis
Finance & Accounting26.7
Rank 130 of 218 measured models
Strategy & Ops29.3
Rank 117 of 218 measured models
Legal27.6
Rank 131 of 218 measured models
Engineering23.9
Rank 146 of 217 measured models
Economics31.6
Rank 145 of 225 measured models
Standard Benchmarks
GPQA Diamond84.5%
Rank 144 of 619 measured models
HLE14.0%
Rank 268 of 664 measured models
SciCode40.0%
Rank 173 of 231 measured models
AA-LCR77.3%
Rank 130 of 575 measured models
Terminal-Bench 2.167.4%
Rank 80 of 244 measured models
Terminal-Bench v42.5%
Rank 109 of 218 measured models
τ-Bench Banking32.2%
Rank 60 of 211 measured models
Models at a similar level
The closest models on the Intelligence Index, one configuration per family.

Preferences and analytics

Rybbit: visits, performance, clicks, copies, forms and session replay (masked inputs), with your permission. Refusing does not limit the site.

Privacy

Allow navigation analytics?

With your permission, Rybbit measures visits, performance and interactions (clicks, copies, forms), including session replay, to improve BenchSift. Input values are masked in replays. Refusing does not limit the site.

Your choice is stored for 180 days. Change it at any time using “Privacy preferences” in the footer. Withdrawal reloads the page.

No choice stored: Rybbit is disabled.

Privacy