Qwen3.8-Max 评测指标直逼Fable5 Max和Qwen3.8-27B模型权重预计下周开源

fengchris 2026-08-03 10:10 1

官方博客：https://qwen.ai/blog?id=qwen3.8

完整性能结果

	Opus4.8	Fable5	GPT5.6 Sol (max)	Qwen3.7-Max	Qwen3.8-Max
Coding Agent
Terminal Bench 2.1	84.6	84.6	88.8	74.5	86.6
SWE-bench Pro	69.2	80.0	64.6	60.6	67.7
DeepSWE 1.1	59.0	70.0	73.0	21.6	56.6
NL2Repo-Bench	69.4	--	--	47.2	55.9
FrontierSWE	70.0	88.8	--	40.7	73.5
MLS-Bench-Lite	42.8	49.9	46.2	31.7	41.0
PaperBench	80.3	88.8	90.5	64.8	93.0
AndroidBench	69.8	84.5	74.0	56.5	70.2
QwenSWEBench	84.0	86.3	73.5	63.4	80.7
QwenQoderBench	62.7	63.1	53.8	36.8	58.4
QwenReactBench	1694	1770	1564	1538	1724
QwenSVGBench	1648	1690	1758	1499	1713
General Agent
CoWorkBench	72.3	75.9	71.5	64.6	74.8
WorkSpaceBench	66.8	68.7	65.6	61.4	67.7
JobBench	48.4	57.4	45.4	31.3	53.4
SkillsBench	65.1	70.9	73.5	61.2	70.2
Agents’ Last Exam (Pass / Score)	27.0 / 45.1	-- / –	30.6 / 53.6	11.8 / 31.1	27.0 / 52.4
Automation-Bench (Pass@1)	27.2	29.1	29.7	14.2	27.3
Toolathlon Verified (Pass@1)	76.2	77.9	74.9	49.7	72.5
WideSearch	72.9	81.2	--	75.2	81.9
HLE w/ tools	57.9	64.5	58.0	53.5	56.2
General Capabilities
GPQA Diamond	92.0	92.6	94.1	92.4	92.6
HLE	45.7	53.3	47.2	41.4	43.6
IFBench	62.2	63.5	72.7	79.1	82.8
$OneMillion-Bench (expert score)	41.8	55.9	53.8	44.4	52.5
HealthBench	52.4	--	55.3	54.5	60.2
PLawBench	69.6	70.2	72.3	58.9	73.2
PRBench-Legal	52.7	57.6	57.6	48.5	57.6
PRBench-Finance	51.9	55.8	55.5	46.8	58.3
MRCR v2 256K (8-needle)	83.2	--	93.8	86.7	92.9
LongBench v2	69.1	--	67.1	65.3	66.3

Opus4.8	Fable5	Gemini3.1-Pro	GPT5.6-Sol	Qwen3.7-Plus	Qwen3.8-Max
Multimodal Reasoning
MMMU-Pro	75.6	81.2	80.5	83.0	79.0
MathVision	87.1 / 97.1	92.7 / 98.6	87.4 / 95.7	90.8 / 97.8	90.3 / –
BabyVision	28.4 / 81.2	42.5 / 90.5	55.9 / 68.3	65.5 / 88.9	64.7 / 70.4
HLE-VL (w/ Tools)	--	--	43.9	51.2	25.6
ZeroBench (Pass@5)	17.0 / 34.0	20.0 / 46.0	17.0 / 23.0	22.0 / 35.0	19.0 / 19.0
ZeroBench-Sub	31.1	37.1	36.5	46.7	41.0
LogicVista	76.7	85.7	82.6	89.7	84.3
HiPhO	69.3	78.6	85.4	86.8	84.1
PhyX	54.2	71.7	79.4	79.1	80.0
SLAKE	75.9	86.6	82.9	85.1	83.2
MedXpertQA-MM	71.7	80.0	80.7	81.5	71.0
PMC-VQA	59.2	63.2	62.5	62.3	63.4
Visual Agent & Coding
OSWorld-Verified	83.4	85.0	76.2	83.2	73.3
OSWorld 2.0	20.6 / 54.8	-- / 66.1	7.8 / 30.6	-- / 62.6	2.8 / 21.5
ScreenSpot Pro	82.3	87.3	68.1	81.3	79.0
WebArena-Verified	67.9	71.3	64.3	69.7	55.3
AndroidWorld	75.0	88.8	70.7	77.6	81.0
MobileWorld	67.5	85.5	58.1	76.9	51.2
ClawEval-MM	73.3 / 73.8	81.2 / 77.5	50.5 / 55.2	81.2 / 78.9	57.4 / 60.1
Vision2Web	62.4	70.5	--	62.1	42.1
QwenBlenderBench	62.4	69.5	23.0	68.6	41.5
Parametric CAD Bench	85.1	87.5	73.5	86.2	73.8
RecreationBench	48.0	56.1	16.2	47.6	30.2
PresentBench	80.9	79.8	55.4	82.9	65.7
Document & Office Intelligence
CharXiv (RQ)	78.5 / 89.9	87.9 / 93.5	84.4 / 89.9	85.1 / 89.1	85.8 / 85.9
OmniDocBench 1.5	86.5	89.5	90.0	86.7	91.4
OCR-Bench-V2 (EN/ZH)	53.9 / 55.3	65.3 / 58.1	64.6 / 58.2	69.0 / 57.3	70.7 / 67.1
CC-OCR-Bench-V2	60.3	72.4	68.9	68.0	72.7
MTVQA-Test	48.1	41.6	54.3	52.7	51.2
MADQA	86.8	86.0	81.1	87.8	87.1
QwenVisualOffice	34.5	32.4	39.6	29.5	32.4
Real-World & Spatial Understanding
RealWorldQA	76.6	85.9	83.5	83.7	86.9
ERQA	57.2	70.0	68.0	70.0	69.8
LingoQA	73.8	77.4	66.8	72.6	83.4
SURDS	62.2	79.4	64.0	63.0	77.2
Visual Perception & Grounding
SimpleVQA	67.3	73.4	73.1	66.6	70.3
WorldVQA	33.9	53.5	54.0	45.1	43.9
MMStar	76.7	80.5	84.0	82.5	83.2
PerceptionBench	47.2	57.2	56.2	59.7	51.1
CountQA	41.3	63.1	72.8	68.6	77.0
RefAdv-S	61.7	68.6	71.9	69.2	73.0
Dense200	20.8	31.1	69.7	55.3	60.7
COCO	50.7	56.4	72.4	61.2	74.2
VisFactor	30.1	54.5	39.8	62.8	42.8
VLMsAreBiased	43.8	61.2	74.1	59.8	36.6
Video Intelligence & Agents
VideoMME (w/ Sub.)	85.4	--	86.7	89.5	88.0
VideoMME v2 (w/ Sub.)	49.0	52.2	66.9	71.1	59.7
VideoMMMU	75.3	81.2	85.3	85.0	85.4
MMVU	67.4	72.0	77.9	81.2	76.6
MLVU (M-Avg)	53.4	--	84.7	87.6	87.4
TVBench	61.5	--	73.0	83.2	78.2
LVBench	67.3	--	75.1	78.8	76.2
LVBench (w/ Mem.)	84.3	90.1	--	84.2	74.5
EgoLife (w/ Mem.)	78.3	82.3	--	70.8	68.8
VideoDR (w/ Search)	65.6	77.1	--	71.3	41.0

与K3的对比，K3在编程方面还是略胜，Qwen胜在多模态：

PS:不知道token plan的优惠会持续多久

果然正式版优惠没了按照API价格换算积分只有夜间5折用不起了

最新回复 (19)

DeH40 Sung 08-03 10:12

1楼

max 这次竟然开源了？我记得 max 很久不开源了
calendar 08-03 10:13

2楼

为什么不放测评啊，qwen 不是说擅长刷榜吗 ^-^
F00XXX 08-03 10:13

3楼

这模型现在进化到什么水平了，刚出的时候在web上试了一下很一般啊。
fengchris 楼主 08-03 10:13

4楼

Max从没开过源吧

K3让阿里老实了被迫开源不然开源宝座坐不了了
源程 08-03 10:16

5楼

啥情况，阿里的qwen3.8 max竟然开源，这没记错应该是多模态模型呀。
HeriX 08-03 10:21

6楼

居然是2.4T的大模型吗, 但是感觉能力有点对不上他这个规模
dvdbv 08-03 10:21

7楼

qwen小模型开源越来越少，每次模型发布都没什么存在感了，就像中规中矩的同学连最后的特长都很少玩了一样
Integer 08-03 10:23

8楼

3.7就很多没开源了，阿里想的挺好想搞闭源模型挣钱，可是能力不行啊
Payaso 08-03 10:23

9楼

qwen最新系列的模型对标GPT哪个版本啊，感觉除了小参数模型本地部署外，存在感好低
fengchris 楼主 08-03 10:26

10楼

QWEN3.8-27B的要开源了再不开小模型要让gemma-4占全了
fengchris 楼主 08-03 10:26

11楼

Max可是对标Sol跟Fable的
dvdbv 08-03 10:28

12楼

还是太大了，以前各种尺寸都有的，怀念^-^
a3members 08-03 10:32

13楼

看这个代码分数好像也不咋地 opus 4.8水平
利陆都2253 08-03 10:32

14楼

期待 qwen 这次能超越 Fable 5 和 GPT 5.6 Sol，毕竟 Fable 5 已经发布快 2 个月了
b9348 08-03 10:34

15楼

不是我黑qwen, 但上次preview38也是高分, 结果实际上发和测是两个模型, 我是没见过站里有人吹的,只见过有人黑的, 各种说qwen3.8不行
fengchris 楼主 08-03 10:35

16楼

上次说是测试的时候放错了权重希望这次正式版能名副其实了

不过Qwen一向都有点过度刷分能超过GLM-5.2就不错了
troubleman 08-03 10:36

17楼

preview这次是发正式版吗？我记得之前一直是preview
a12908 08-03 10:37

18楼

qwen每次的官方跑分水分都大得很在站内基本无人问津不像K3和v4都能把站干卡咯

到时候看看这所谓的直逼肥波5max到底能不能打过v4f吧
jerry wu 08-03 10:37

19楼

人工智能大会一把手定调了。。不管你想不想，都得开