Skip to content

Qwen3.8-27B

AlibabaOpen weightsReasoningVerifiedReleased Aug 5, 2026Input: text, image, video
Compare with GPT-5.2Source scores ↗OpenRouter ↗
Overall score
59.6
Rank 58 of 222 · 53.7–65.5
Context window
262K
262,000 tokens
Price per 1M tokens
$0.040 in · $1.35 out
Source: OpenRouter
Blended price
$0.37
3:1 input:output tokens
Output speed
—
Not measured yet

Category scores

Scores per capability area with the model's rank among ranked models in that category.

  1. 1Instruction following#4186.9
  2. 2Multimodal#1384.4
  3. 3Agentic & tool use#2181.7
  4. 4Reasoning#781.3
  5. 5Coding#5259.2
  6. 6Knowledge#7952.2
Higher is better. #N is the model's rank in that category, where it is eligible to be ranked.

Benchmark results (41)

As reported in system cards, papers and public evaluations. Position compares this model with every other model reporting the same benchmark.

Multimodal

CharXiv90.2#7 of 42
ParseBench70.79#8 of 37
ExtractBench (LlamaIndex)(lower is better)89.75#15 of 24
MathVision90.0#7 of 20
RealWorldQA85.9#5 of 16
ERQA65.5#9 of 16
CharXiv w/o tools83.7#7 of 14
OmniDocBench 1.591.1#4 of 11
BabyVision65.7#5 of 8
MathVision w/ Python94.6#5 of 5
BabyVision w/ Python85.6#5 of 5
Vision2Web62.9#3 of 4

Agentic & tool use

Terminal-Bench 2.173.0#38 of 68
OSWorld-Verified84.3#4 of 35
JobBench33.4#22 of 34
Agents' Last Exam42.9#5 of 17
AndroidWorld81.9#4 of 8
CoWorkBench70.7#4 of 4
WebArena-Verified64.8#3 of 4

Coding

SWE-bench Pro61.7#23 of 69
DeepSWE42.2#48 of 50
LiveCodeBench v690.3#6 of 46
NL2Repo42.3#21 of 31
Bug Hunt Bench15.0#23 of 23
VulcanBench v382.6#10 of 14

Knowledge

GPQA89.2#30 of 117
HLE30.8#41 of 69
HLE w/o tools30.8#31 of 43

Instruction following

IFBench79.5#12 of 42

Pricing by source

USD per 1M tokens.

SourceInputOutputCached in
List price (OpenRouter)$0.040$1.35—
OpenRouter qwen/qwen3.8-27b$0.040$1.35$0.018

Qwen publishes Qwen3.8-27B under Apache-2.0 for self-hosting and has not published a first-party hosted token rate for the exact model. BenchLM represents the open-weight row as self-host/free-per-token before infrastructure costs. Qwen Cloud says hosted access is coming soon, so the pricing row does not inherit Qwen3.8 Max rates.

Price history

  • Oct 17, 2026: $0.24 / $2.20
  • Oct 20, 2026: $0.040 / $1.35

Arena ratings

Human-preference ratings from LMArena. Best-rated variant shown.

Text1441#75 · 23.3K votes
WebDev (coding)1593#31 · 16.4K votes
Vision1270#44 · 7.6K votes

Open weights

From the Hugging Face repository.

Parameters
28B
Licence
apache-2.0
Downloads (30 days)
7.6M

Speed

Measured by us through OpenRouter (median of trials).

Not measured yet.

Related models

Same family

Similar score