Qwen 3.6 Plus

Qwen 3.6 Plus ranks first on 6 of the 158 benchmarks with five or more models scored.

Qwen Released March 31st, 2026 Family Qwen3.6 The lab's model card Compare with others

160

Benchmarks with a score

Ranked first, out of five or more models

In the top three

Qwen Mar 2026

Qwen 3.6 Plus

Closed weights Flagship

148.9 #74 of 256

Epoch Capabilities Index

Self-Reported Index 151.1 #87 of 404

SWE-bench Verified 57.9% #29 /32
MMMU-Pro 78.8% #22 /90
IFBench 74.2% #26 /80

Where it ranks

Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.

Knowledge and language

Benchmark	Score	Rank	Source
MMLU-Pro	88.5%	#3 of 167	Lab model card
C-Eval	93.3%	#2 of 46	Lab model card
MMLU-Redux	94.5%	#3 of 45	Lab model card
IFEval	94.3%	#10 of 135	Lab model card
SuperGPQA	71.6%	#3 of 35	Lab model card
BullshitBench V2	72%	#18 of 129	Official leaderboard
MMMLU	89.5%	#11 of 71	Lab model card
MMLU-ProX	84.7%	#4 of 23	Lab model card
LMArena (Elo)	1,443	#72 of 335	Official leaderboard
INCLUDE	85.1%	#7 of 31	Lab model card
LongBench v2	62%	#8 of 35	Lab model card
Humanity's Last Exam	50.6%	#38 of 154	Lab model card
AA-LCR (Artificial Analysis Long Context Reasoning)	68.3%	#12 of 38	Lab model card
IFBench	74.2%	#26 of 80	Lab model card
CL-bench	20.3%	#8 of 24	Official leaderboard
MAXIFE	88.2%	#4 of 11	Lab model card
SimpleVQA	67.3%	#12 of 30	Lab model card
WideSearch	74.3%	#8 of 19	Lab model card
SimpleQA Verified	44.1%	#46 of 94	Epoch AI run
Global PIQA	89.8%	#7 of 14	Lab model card
WMT24++	84.3%	#8 of 15	Lab model card
NOVA-63	57.9%	#6 of 11	Lab model card
OpenAI MRCR v2	41%	#26 of 47	Official leaderboard
Extended NYT Connections	60.3%	#66 of 103	Official leaderboard
DesignArena ASCII Art	1,110	#63 of 77	Official leaderboard
Lech Mazur Creative Writing	-2.33	#42 of 50	Official leaderboard

Math and science

Benchmark	Score	Rank	Source
PolyMath	77.4%	#3 of 24	Lab model card
HMMT Nov 2025	94.6%	#4 of 28	Lab model card
HMMT 2025	96.7%	#11 of 54	Lab model card
GPQA Diamond	88.4%	#60 of 229	Epoch AI run
AIME 2026	95.3%	#10 of 38	Lab model card
OTIS Mock AIME 2024-2025	93.3%	#57 of 212	Epoch AI run
FrontierMath	26.2%	#19 of 55	Official leaderboard
FrontierMath Tier 4	8.33%	#21 of 55	Epoch AI run
IMO-AnswerBench	83.8%	#17 of 37	Lab model card
DTBench	81.9%	#83 of 180	Official leaderboard
ResearchClawBench	18.0	#13 of 23	Official leaderboard
HMMT 2026	87.8%	#12 of 19	Lab model card
FrontierMath Tiers 1-3 v2	38.3%	#59 of 86	Official leaderboard

Coding

Benchmark	Score	Rank	Source
LiveCodeBench v6	87.1%	#15 of 90	Lab model card
DesignArena Website	1,235	#65 of 151	Official leaderboard
DesignArena Code	1,227	#63 of 143	Official leaderboard
DesignArena Game Dev	1,202	#63 of 143	Official leaderboard
DesignArena Data Visualization	1,224	#53 of 118	Official leaderboard
SWE-bench Multilingual	73.8%	#33 of 73	Lab model card
DesignArena UI Component	1,223	#65 of 139	Official leaderboard
OpenHands Index	52.9%	#16 of 33	Official leaderboard
DesignArena 3D Design	1,186	#66 of 136	Official leaderboard
QwenWebBench	1,502	#2 of 4	Lab model card
WebDev Arena	1,461	#53 of 105	arena.ai, via Epoch AI
SWE-Bench Pro	56.6%	#42 of 81	Lab model card
Code Arena	1,461	#68 of 128	Official leaderboard
EvoCode-Bench	29.4%	#7 of 13	Official leaderboard
DesignArena SVG	1,152	#56 of 103	Official leaderboard
ALE-Bench	1,273	#77 of 126	Official leaderboard
Vibe Code Bench	25.6%	#81 of 120	Official leaderboard
EvoCode-Bench v2	9.70%	#9 of 13	Official leaderboard
DesignArena Video to Website	1,170	#14 of 19	Official leaderboard
Image-to-WebDev Arena	1,473	#41 of 55	Official leaderboard
QwenVision2Code	1,522	#4 of 5	Official leaderboard
RoadmapBench	12.2%	#9 of 11	Official leaderboard
Code Migration	11.1%	#64 of 78	Official leaderboard
NL2Repo-Bench	37.9%	#25 of 30	Lab model card
QwenSVG	1,432	#6 of 7	Official leaderboard
SWE-bench Verified	57.9%	#29 of 32	Epoch AI run
FrogsGame post-training task	0%	#17 of 17	Official leaderboard
FrontierSWE v1	19.9%	#17 of 17	Official leaderboard
AutoLab	0.27	#10 of 10	Official leaderboard
QwenWebDev	1,500	#6 of 6	Official leaderboard

Agents and tools

Benchmark	Score	Rank	Source
DeepPlanning	41.5%	#2 of 10	Lab model card
DeepTerminalBench	32%	#2 of 9	Official leaderboard
SnakeBench	28	#53 of 188	Official leaderboard
Terminal-Bench 2.0	61.6%	#18 of 63	Lab model card
VITA-Bench	44.3%	#5 of 16	Lab model card
Pencil Puzzle Bench	10%	#17 of 53	Official leaderboard
QwenClawBench	57.4%	#4 of 12	Official leaderboard
LiveClawBench	70.6%	#6 of 17	Official leaderboard
tau3-bench	70.7%	#5 of 14	Lab model card
Toolathlon	39.8%	#19 of 50	Official leaderboard
ClawProBench	60.2%	#22 of 49	Official leaderboard
PinchBench 2.0	72.5%	#29 of 57	Official leaderboard
ScreenSpot-Pro	68.2%	#17 of 32	Lab model card
MCP Atlas	74.1%	#26 of 46	Lab model card
Claw-Eval	75%	#14 of 24	Official leaderboard
ToolcallFormatIFBench	83.3%	#3 of 5	Official leaderboard
Long-Horizon-Terminal-Bench (LHTB)	31.3%	#16 of 25	Official leaderboard
Claw-Eval Pass^3	58.7%	#9 of 13	Lab model card
MCPMark	48.2%	#7 of 10	Lab model card
OSWorld-Verified	62.5%	#29 of 41	Lab model card
TERMS-Bench	0.60	#8 of 11	Official leaderboard
PinchBench	63.3%	#35 of 48	Official leaderboard
MobileGym	45.7%	#3 of 4	Official leaderboard
SkillsBench	45.7%	#11 of 14	Lab model card
Mystery Game Puzzles	12%	#68 of 84	Epoch AI run
AutomationBench	4.30%	#34 of 42	Official leaderboard
Chess Puzzles	16%	#77 of 93	Epoch AI run
Claw-Eval-Live	50.5%	#11 of 12	Official leaderboard

Professional work

Benchmark	Score	Rank	Source
MortgageTax	68%	#20 of 87	Official leaderboard
TaxEval v2	74.7%	#31 of 129	Official leaderboard
LegalBench	84.2%	#37 of 138	Official leaderboard
Vals Finance Agent	54.6%	#15 of 49	Official leaderboard
Workspace-Bench-Lite	55.6%	#3 of 9	Official leaderboard
CorpFin v2	61.9%	#51 of 119	Official leaderboard
Vals Index v1	53.3%	#16 of 34	Official leaderboard
Vending-Bench 2	$5,115	#34 of 67	Official leaderboard
Clipping Bench	$458	#12 of 19	Official leaderboard
FoodTruck Bench	$7,668	#40 of 59	Official leaderboard
YC-Bench	$787,999	#40 of 59	Official leaderboard
MedScribe	77%	#66 of 97	Official leaderboard
Harvey LAB (Legal Agent Benchmark)	1.25%	#55 of 80	Official leaderboard
FORTE	62.5%	#15 of 21	Official leaderboard
MedCode	36.9%	#71 of 96	Official leaderboard
PhysicianBench	13.7%	#10 of 13	Official leaderboard
Vals Finance Agent v2	40.9%	#63 of 80	Official leaderboard
Vals Index v2	32%	#52 of 65	Official leaderboard
Excel Modeling Benchmark	32.9%	#63 of 76	Official leaderboard
MBABench	14.7	#5 of 6	Official leaderboard
Legal Research Bench	14.9%	#78 of 93	Official leaderboard
Agents' Last Exam	8.55%	#27 of 31	Official leaderboard
Tax Agent Bench	7.68%	#74 of 84	Official leaderboard
CaseLaw v2	51.5%	#49 of 53	Official leaderboard
E-Commerce Bench	0.47	#18 of 19	Official leaderboard
TRLawBench	6.97	#33 of 34	Official leaderboard

Vision, audio and robotics

Benchmark	Score	Rank	Source
MMMU	86%	#1 of 103	Lab model card
DynaMath	88%	#1 of 23	Lab model card
V*	96.9%	#1 of 17	Lab model card
ODinW-13	51.8%	#1 of 13	Lab model card
We-Math	89%	#1 of 13	Lab model card
MMStar	83.3%	#3 of 36	Lab model card
MMLongBench-Doc	62%	#2 of 19	Lab model card
AI2D	94.4%	#6 of 52	Lab model card
CC-OCR-Bench-V2	76.5%	#2 of 15	Official leaderboard
CountBench	97.6%	#4 of 28	Lab model card
CC-OCR	83.4%	#2 of 14	Lab model card
Video-MME	87.8%	#8 of 51	Lab model card
TIR-Bench	61.6%	#1 of 6	Lab model card
RealWorldQA	85.4%	#8 of 38	Lab model card
VisFactor	38.2%	#6 of 26	Official leaderboard
MMMU-Pro	78.8%	#22 of 90	Lab model card
MATH-Vision (MathVision)	88%	#13 of 51	Lab model card
MLVU	86.7%	#6 of 22	Lab model card
ERQA	65.7%	#12 of 36	Lab model card
CharXiv	81.5%	#26 of 75	Lab model card
Video-MMMU	84%	#13 of 36	Lab model card
CiteVQA	17.5%	#10 of 16	Official leaderboard
AgentWorldBench	50.8	#11 of 13	Official leaderboard
BVB	53.9	#42 of 45	Official leaderboard
VoxelBench	1,263	#54 of 56	Official leaderboard
QwenWorldBench	47.6%	#5 of 5	Official leaderboard

Safety and security

Benchmark	Score	Rank	Source
nocot-bench (NCRI)	82.2	#53 of 185	Official leaderboard
HelpBench	83%	#4 of 9	Official leaderboard
Human Pathogen Capabilities Test	50.2%	#46 of 86	Official leaderboard
Molecular Biology Capabilities Test	47.2%	#52 of 90	Official leaderboard
World-Class Bio	43.6%	#43 of 74	Official leaderboard
Dictatorship Eval	38.8%	#15 of 20	Official leaderboard

Other

Benchmark	Score	Rank	Source
SAGE (Vals)	44.9%	#42 of 81	Official leaderboard
Conceptual Reasoning Index	45.7	#86 of 151	Official leaderboard
LMCA	33.1%	#85 of 147	Official leaderboard
ACCoRD	59.3	#92 of 152	Official leaderboard
Vals Multimodal Index	51.5%	#25 of 33	Official leaderboard

Its lab's numbers put it at 151.1, 2.3 points above its ECI

The Self-Reported Index fits one score to the 59 capability results Qwen reported for this model, on the scale of the Epoch Capabilities Index. Labs choose which benchmarks to show and how to run them, so read it as the lab's claim, not an independent measurement. The 90% interval runs from 150.1 to 152.1. How we compute it.

CountBench 97.6
V* 96.9
HMMT 2025 96.7
AIME 2026 95.3
HMMT Nov 2025 94.6
MMLU-Redux 94.5
AI2D 94.4
IFEval 94.3
C-Eval 93.3
GPQA Diamond 90.4
Global PIQA 89.8
MMMLU 89.5
We-Math 89
MMLU-Pro 88.5
MAXIFE 88.2
MATH-Vision (MathVision) 88
DynaMath 88
HMMT 2026 87.8
Video-MME 87.8
LiveCodeBench v6 87.1
MLVU 86.7
MMMU 86
RealWorldQA 85.4
INCLUDE 85.1
MMLU-ProX 84.7
WMT24++ 84.3
Video-MMMU 84
IMO-AnswerBench 83.8
CC-OCR 83.4
MMStar 83.3
CharXiv 81.5
SWE-bench Verified 78.8
MMMU-Pro 78.8
PolyMath 77.4
Claw-Eval 74.8
WideSearch 74.3
IFBench 74.2
MCP Atlas 74.1
SWE-bench Multilingual 73.8
SuperGPQA 71.6
tau3-bench 70.7
AA-LCR (Artificial Analysis Long Context Reasoning) 68.3
ScreenSpot-Pro 68.2
SimpleVQA 67.3
OSWorld-Verified 62.5
LongBench v2 62
MMLongBench-Doc 62
Terminal-Bench 2.0 61.6
Claw-Eval Pass^3 58.7
NOVA-63 57.9
SWE-Bench Pro 56.6
ODinW-13 51.8
Humanity's Last Exam 50.6
MCPMark 48.2
SkillsBench 45.7
VITA-Bench 44.3
DeepPlanning 41.5
Toolathlon 39.8
NL2Repo-Bench 37.9

Its lab's numbers next to 6 leaderboards

Where the benchmark's own leaderboard, or Epoch AI's run, also has this model. Gaps mostly come from different harnesses, effort levels and task subsets. Every pair.

Benchmark	Lab says	Board says	Gap	What differs
SWE-bench Verified	78.8% SWE-bench Verified	57.9% Epoch AI run, 2026-04-02	+21 pts	Qwen uses its internal coding scaffold, while Epoch uses its own tool loop and a validated 484-task subset.
GPQA Diamond	90.4% GPQA	88.4% Epoch AI run, 2026-04-02	+2.0 pts
QwenClawBench	57.2% QwenClawBench	57.4% Leaderboard, v1.1	−0.2 pts
Claw-Eval	74.8% Claw-Eval Avg	75% Leaderboard	−0.2 pts
Claw-Eval Pass^3	58.7% Claw-Eval Pass^3	58.8% Leaderboard	Same
Toolathlon	39.8% Tool Decathlon	39.8% Leaderboard	Same

Qwen's models over time

Qwen's models in color, against every other model on the Epoch Capabilities Index, by release date.

Source: Epoch Capabilities Index, Epoch AI (CC BY 4.0)

Its launch post

The benchmarks the lab chose to headline, with the scores it gave.

Qwen3.6-Plus: Towards Real World Agents (blog) Apr 2, 2026

SWE-bench Verified 78.8
SWE-bench Multilingual 73.8
SWE-bench Pro 56.6
Terminal-Bench 2.0 61.6
Claw-Eval Avg 74.8
Claw-Eval Pass^3 58.7
SkillsBench Avg5 45.7
QwenClawBench 57.2
NL2Repo 37.9
QwenWebBench 1501.7
TAU3-Bench 70.7
VITA-Bench 44.3
DeepPlanning 41.5
Tool Decathlon 39.8
MCPMark 48.2
MCP-Atlas 74.1
Best of HLE w/ tool / HLE 50.6
WideSearch 74.3
MMLU-Pro 88.5
MMLU-Redux 94.5
SuperGPQA 71.6
C-Eval 93.3
IFEvalstrict prompt 94.3
IFBench 74.2
AA-LCR 68.3
LongBench v2 62
GPQA 90.4
LiveCodeBench v6 87.1
HMMT Feb 25 96.7
HMMT Nov 25 94.6
HMMT Feb 26 87.8
IMOAnswerBench 83.8
AIME26 95.3
MMMLU 89.5
MMLU-ProX 84.7
NOVA-63 57.9
INCLUDE 85.1
Global PIQA 89.8
PolyMATH 77.4
WMT24++ 84.3
MAXIFE 88.2
MMMU 86
MMMU-Pro 78.8
MathVision 88
We-Math 89
DynaMath 88
RealWorldQA 85.4
MMStar 83.3
SimpleVQA 67.3
OmniDocBench1.5 91.2
CharXiv(RQ) 81.5
MMLongBench-Doc 62
CC-OCR 83.4
AI2D_TEST 94.4
CountBench 97.6
ODinW13 51.8
ERQA 65.7
V* (96.9 / 90.5) 96.9
Best of VideoMME(w sub.) / VideoMME(w/o sub.) 87.8
VideoMMMU 84
MLVU (M-Avg) 86.7
ScreenSpot Pro 68.2
TIR-Bench (61.6 / 43.7) 61.6
OSWorld-Verified 62.5