Qwen3.7-Max

Qwen3.7-Max ranks first on 10 of the 143 benchmarks with five or more models scored.

Qwen Released May 19th, 2026 Family Qwen3.7 The lab's model card Compare with others

144

Benchmarks with a score

Ranked first, out of five or more models

In the top three

Qwen May 2026

Qwen3.7-Max

Closed weights Flagship

154.5 #49 of 256

Epoch Capabilities Index

Self-Reported Index 155.8 #55 of 404

SWE-bench Verified 77.3% #7 /32
IFBench 79.1% #12 /80
SciCode 53.5% #6 /49

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	89.6%	#2 of 167	Lab model card
MMLU-Redux	95%	#1 of 45	Lab model card
MMLU-ProX	87%	#1 of 23	Lab model card
SuperGPQA	73.6%	#2 of 35	Lab model card
IFEval	94.3%	#11 of 135	Lab model card
MAXIFE	89.2%	#1 of 11	Lab model card
HUMAINE	33	#5 of 50	Official leaderboard
LMArena (Elo)	1,474	#35 of 335	Official leaderboard
KINA	49.2%	#4 of 37	Official leaderboard
MMMLU	90.3%	#9 of 71	Lab model card
Phare	75.6%	#8 of 62	Official leaderboard
BullshitBench V2	72%	#19 of 129	Official leaderboard
IFBench	79.1%	#12 of 80	Lab model card
INCLUDE	86.2%	#5 of 31	Lab model card
NOVA-63	59%	#2 of 11	Lab model card
Humanity's Last Exam	53.5%	#30 of 154	Lab model card
MRCR v2	90.4%	#6 of 30	Lab model card
CompassArena	1,055	#2 of 10	Official leaderboard
BullshitBench	52.7%	#26 of 107	Official leaderboard
SimpleQA Verified	55.8%	#25 of 94	Epoch AI run
Wer wird Millionär? LLM Benchmark	87044.0	#21 of 77	Official leaderboard
Global PIQA	91.4%	#4 of 14	Lab model card
Antidote	1,051	#11 of 36	Official leaderboard
Extended NYT Connections	85.1%	#34 of 103	Official leaderboard
WMT24++	85.8%	#5 of 15	Lab model card
DesignArena ASCII Art	1,204	#27 of 77	Official leaderboard
OpenAI MRCR v2	44.5%	#23 of 47	Official leaderboard
EQ-Bench 4	1,100	#27 of 32	Official leaderboard
Lech Mazur Creative Writing	-2.71	#44 of 50	Official leaderboard

Math and science

Benchmark	Score	Rank	Source
PolyMath	86.5%	#1 of 24	Lab model card
HMMT 2026	97.1%	#1 of 19	Lab model card
IMO-AnswerBench	90%	#3 of 37	Lab model card
GPQA Diamond	90.9%	#38 of 229	Epoch AI run
DTBench	92.3%	#37 of 180	Official leaderboard
OTIS Mock AIME 2024-2025	95.6%	#48 of 212	Epoch AI run
SimpleBench	70.4%	#24 of 106	Official leaderboard
MathArena (Apex, IMO, USAMO)	44.5%	#5 of 12	Lab model card
FrontierMath Tier 4 v2	34.1%	#29 of 68	Official leaderboard
ProofBench	26%	#26 of 59	Official leaderboard
FrontierMath Tiers 1-3 v2	64.6%	#38 of 86	Official leaderboard
ResearchClawBench	18.7	#11 of 23	Official leaderboard
CritPt	11.4%	#8 of 13	Lab model card
xbench-ScienceQA	70.8%	#18 of 29	Official leaderboard
NatureBench	7.78%	#13 of 18	Official leaderboard

Coding

Benchmark	Score	Rank	Source
LiveCodeBench	91.6%	#2 of 54	Lab model card
SciCode	53.5%	#6 of 49	Lab model card
QwenSVG	1,608	#1 of 7	Official leaderboard
IOI (Vals)	46.8%	#11 of 55	Official leaderboard
SWE-bench Verified	77.3%	#7 of 32	Epoch AI run
DesignArena Data Visualization	1,274	#29 of 118	Official leaderboard
SWE-bench Multilingual	78.3%	#18 of 73	Lab model card
DesignArena Website	1,268	#38 of 151	Official leaderboard
DesignArena Code	1,266	#40 of 143	Official leaderboard
DesignArena 3D Design	1,248	#39 of 136	Official leaderboard
DesignArena Game Dev	1,255	#42 of 143	Official leaderboard
DesignArena UI Component	1,264	#43 of 139	Official leaderboard
SWE-Bench Pro	60.6%	#27 of 81	Lab model card
DesignArena SVG	1,207	#35 of 103	Official leaderboard
FrontierCS	61.9	#5 of 14	Official leaderboard
WebDev Arena	1,517	#39 of 105	arena.ai, via Epoch AI
Code Arena	1,515	#56 of 128	Official leaderboard
Claw-SWE-Bench	73.1%	#7 of 16	Official leaderboard
DesignArena Agentic Overall	1,171	#36 of 74	Official leaderboard
QwenWebDev	1,568	#3 of 6	Official leaderboard
DesignArena Frontend Web App	1,189	#27 of 53	Official leaderboard
DesignArena Full Stack	1,172	#25 of 49	Official leaderboard
NL2Repo-Bench	47.2%	#16 of 30	Lab model card
Vibe Code Bench	47.7%	#68 of 120	Official leaderboard
Monthly-SWEBench	51%	#7 of 12	Official leaderboard
EvoCode-Bench v2	11.9%	#8 of 13	Official leaderboard
DesignArena React Native	1,152	#35 of 56	Official leaderboard
DesignArena Godot Game Dev	1,169	#15 of 24	Official leaderboard
HarnessDev	52.8%	#4 of 6	Official leaderboard
DesignArena Android (Kotlin)	1,148	#32 of 46	Official leaderboard
Code Migration	13.1%	#60 of 78	Official leaderboard
DesignArena Agentic Game Dev	1,106	#21 of 27	Official leaderboard
ContextBench	18.5%	#7 of 9	Official leaderboard
MLS-Bench	31.7%	#12 of 15	Official leaderboard
AndroidBench	54.2%	#27 of 32	Official leaderboard
GBAEval	0.41%	#20 of 21	Official leaderboard
QwenQoderBench	36.8%	#5 of 5	Official leaderboard
QwenReactBench	1,538	#5 of 5	Official leaderboard
QwenSVGBench	1,499	#5 of 5	Official leaderboard
QwenSWEBench	63.4%	#5 of 5	Official leaderboard

Agents and tools

Benchmark	Score	Rank	Source
BFCL V4	75%	#1 of 30	Lab model card
PinchBench 2.0	92.5%	#2 of 57	Official leaderboard
SnakeBench	34	#12 of 188	Official leaderboard
Terminal-Bench 2.0	69.7%	#5 of 63	Lab model card
VITA-Bench	47.9%	#2 of 16	Lab model card
Pencil Puzzle Bench	40%	#7 of 53	Official leaderboard
QwenClawBench	64.3%	#1 of 6	Lab model card
CEO-Bench	$365,346	#5 of 19	Official leaderboard
MCPMark	60.8%	#3 of 10	Lab model card
Mystery Game Puzzles	32%	#32 of 84	Epoch AI run
SkillsBench	59.2%	#6 of 14	Lab model card
MCP Atlas	76.4%	#20 of 46	Lab model card
PACT	1,511	#14 of 26	Official leaderboard
Chess Puzzles	27%	#55 of 93	Epoch AI run
TUA-Bench	44.9%	#8 of 12	Official leaderboard
Claw-Eval	65.2%	#9 of 13	Lab model card
Long-Horizon-Terminal-Bench (LHTB)	29.6%	#18 of 25	Official leaderboard
RuneBench	4.07	#56 of 76	Official leaderboard
Agent Arena	-7.15%	#45 of 56	Official leaderboard
EBR-bench	9.52%	#26 of 31	Epoch AI run
Harbor-Index	4.90%	#8 of 9	Official leaderboard
BabaIsBench	25%	#31 of 32	Official leaderboard

Professional work

Benchmark	Score	Rank	Source
MerchantBench	59.5	#1 of 8	Official leaderboard
TaxEval v2	75.3%	#19 of 129	Official leaderboard
LegalBench	84.9%	#28 of 138	Official leaderboard
SpreadsheetBench	87%	#2 of 6	Lab model card
CorpFin v2	63.7%	#42 of 119	Official leaderboard
FoodTruck Bench	$27,827	#25 of 59	Official leaderboard
Excel Modeling Benchmark	57%	#42 of 76	Official leaderboard
MedScribe	79.4%	#58 of 97	Official leaderboard
Vals Index v2	44.8%	#39 of 65	Official leaderboard
DesignArena Python-PPTX Slides	1,147	#22 of 36	Official leaderboard
Harvey LAB (Legal Agent Benchmark)	1.67%	#52 of 80	Official leaderboard
Vals Finance Agent v2	47.8%	#52 of 80	Official leaderboard
MedCode	38.8%	#65 of 96	Official leaderboard
DesignArena HTML Slides	1,054	#19 of 28	Official leaderboard
Legal Research Bench	25.5%	#65 of 93	Official leaderboard
Tax Agent Bench	15.8%	#61 of 84	Official leaderboard
DesignArena PDF	1,142	#27 of 37	Official leaderboard
YC-Bench	$629,583	#44 of 59	Official leaderboard
DesignArena Spreadsheet	1,086	#27 of 35	Official leaderboard
Agents' Last Exam	11.8%	#24 of 31	Official leaderboard
EnterpriseBench: CoreCraft	26.2%	#40 of 51	Official leaderboard
E-Commerce Bench	1.65	#15 of 19	Official leaderboard
PolicyBench	76.9%	#41 of 50	Official leaderboard
CoWorkBench	64.6%	#5 of 5	Official leaderboard

Vision, audio and robotics

Benchmark	Score		Rank	Source
VoxelBench	1,609		#33 of 56	Official leaderboard
QwenWorldBench	57.3%		#2 of 2	Lab model card

Safety and security

Benchmark	Score	Rank	Source
BioTIER Permit	99.9%	#7 of 69	Official leaderboard
SecCodeBench	71.1%	#1 of 6	Official leaderboard
nocot-bench (NCRI)	85.8	#44 of 185	Official leaderboard
BrowseSafe++	89%	#4 of 15	Official leaderboard
Molecular Biology Capabilities Test	51.8%	#32 of 90	Official leaderboard
Human Pathogen Capabilities Test	56.8%	#31 of 86	Official leaderboard
World-Class Bio	50.2%	#28 of 74	Official leaderboard
BioTIER Refuse	32.8%	#41 of 69	Official leaderboard
DrugDiscoveryBench	29.3%	#18 of 20	Official leaderboard

Other

Benchmark	Score	Rank	Source
LMCA	44%	#46 of 147	Official leaderboard
Conceptual Reasoning Index	56.9	#48 of 151	Official leaderboard
ACCoRD	65.3	#73 of 152	Official leaderboard

Its lab's numbers put it at 155.8, 1.2 points above its ECI

The Self-Reported Index fits one score to the 33 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 154.3 to 157.3. How we compute it.

Its lab's numbers next to 5 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
SWE-bench Verified	80.4% SWE-Verified	77.3% Epoch AI run	+3.1 pts
CoWorkBench	67.2% CoWorkBench	64.6% Leaderboard	+2.6 pts
GPQA Diamond	92.4% GPQA Diamond	90.9% Epoch AI run	+1.5 pts
QwenSVG	1,608 QwenSVG	1,608 Leaderboard	Same
QwenWebDev	1,568 QwenWebDev	1,568 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.

Qwen3.7: The Agent Frontier (blog) May 20, 2026