Inference
EN
Alibaba

Qwen3.6 Plus

Qwen3.6 Plus is a reasoning model from Alibaba. 58 benchmarks count toward its index, in 8 categories.

  • Closed weights
  • Reasoning
  • Input: text · image · video
  • Released 2026-04
RankedEnough results for an index.
IndexAll categories, out of 100.54.6±2.1Median 51.0
CoverageIndex weight that has results.100%
SpeedOutput tokens per second.44/sMedian 67/s
First tokenSeconds to the first output token.1.1 sMedian 1.3 s
Input / 1MUS dollars per 1M input tokens.$0.325Median $0.60
Output / 1MUS dollars per 1M output tokens.$1.95Median $2.55
ContextMaximum tokens in one request.1MMedian 262K
EloLMArena rating and rank.1436±4 #86Median 1422

Details

From OpenRouter
API IDThe model ID on OpenRouter.
qwen/qwen3.6-plus
Max outputMaximum output tokens per request.
66K
Knowledge cutoffLast date of training data.
—
ToolsTool calls on OpenRouter.
Yes
JSON outputOutput follows a given JSON schema.
Yes
Cache read / 1MUS dollars per 1M cached tokens.
—
WeightsLink to the published weights.
—
RetiresDate OpenRouter removes the model.
—

Where to run

APIs that serve this model.
2 providers
ProviderTypeMaker, cloud, host or router.InputOutputContext
OpenRouterRouter$0.325$1.951M
Together AIHost$0.50$3.001M

US dollars per 1M tokens, list price. From models.dev (opens in a new tab) (MIT).

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.15 counted55.0103 / 218
CodingCode writing and repair.8 counted56.289 / 209
ReasoningLogic problems and puzzles.6 counted51.2107 / 227
MultimodalTasks with images and text.6 counted56.151 / 84
KnowledgeFacts and expert knowledge.9 counted54.1102 / 268
MultilingualTasks in many languages.1 counted54.1—
InstructionTasks with strict prompt rules.3 counted52.435 / 63
MathMath problems.10 counted54.965 / 168

Results

58 counted
BenchmarkThe benchmark name.ResultThe score from the source.PlacePlace among results on this benchmark.
Massive Multitask Language Understanding ProfessionalKnowledge88.5%3 / 75
τ²-Bench Tool-Agent-User EvaluationAgentic97.7%9 / 149
V*Multimodal96.9%1 / 14
Massive Multi-discipline Multimodal UnderstandingMultimodal86.0%3 / 28
Instruction-Following EvalInstruction94.3%8 / 70
AIMEMath94.6% ±0.613 / 95
τ³-Bench Tool-Agent-User EvaluationAgentic70.7%5 / 32
Harvard-MIT Mathematics Tournament November 2025Math94.6%2 / 12
C-EvalKnowledge93.3%2 / 12
LiveCodeBenchCoding86.0% ±1.024 / 142
Artificial Analysis IFBenchInstruction75.2%25 / 146
MMLU ProKnowledge87.7% ±0.324 / 135
Graduate-Level Google-Proof Q&AKnowledge90.4%27 / 141
Harvard-MIT Mathematics Tournament February 2025Math96.7%3 / 15
MMLU-ReduxKnowledge94.5%4 / 19
SuperGPQA: Scaling LLM Evaluation Across 285 Graduate DisciplinesKnowledge71.6%5 / 22
Software Engineering Benchmark VerifiedCoding78.8%26 / 114
MMLU-ProXMultilingual84.7%4 / 17
GPQA diamond · Epoch AIKnowledge88.4% ±2.354 / 217
OTIS Mock AIME 2024-2025Math93.3% ±3.851 / 200
Artificial Analysis GPQA DiamondKnowledge88.2%52 / 199
GPQA Diamond · Vals AIKnowledge87.4% ±1.738 / 135
LiveCodeBench v6Coding87.1%17 / 59
Artificial Analysis Long Context ReasoningReasoning78.3%63 / 217
MMMU ProMultimodal84.2% ±0.927 / 92
VITA-BenchAgentic44.3%4 / 13
LongBench v2Reasoning62.0%8 / 25
FrontierMath-2025-02-28-PrivateMath26.2% ±2.623 / 69
Artificial Analysis MMMU-ProMultimodal78.0%39 / 113
Terminal-Bench 2.0 · Vals AIAgentic44.9% ±5.324 / 66
FrontierMath-Tier-4-2025-07-01-PrivateMath8.3% ±4.021 / 55
AIME 2026Math95.3%13 / 34
Terminal-Bench 2.0 · BenchLMAgentic61.6%16 / 41
Massive Multi-discipline Multimodal Understanding ProMultimodal78.8%21 / 53
QwenClawBenchAgentic57.2%4 / 10
Gert Labs Composite Game BenchmarkAgentic50.6%21 / 52
Artificial Analysis Humanity's Last ExamKnowledge27.8%92 / 222
LiveBench CodingCoding78.2%29 / 66
Harvard-MIT Mathematics Tournament February 2026Math87.8%12 / 27
Critical Physics TasksReasoning2.9%98 / 214
MathVisionMultimodal88.0%11 / 24
Instruction Following BenchmarkInstruction75.8%28 / 57
SimpleQA VerifiedKnowledge44.1% ±1.645 / 91
CharXiv ReasoningMultimodal81.5%27 / 49
GDPval-AA normalizedAgentic24.5%74 / 133
WideResearchAgentic74.3%9 / 16
SWE-bench ProCoding56.6%47 / 83
Artificial Analysis Omniscience AccuracyKnowledge26.4%95 / 164
SWE-benchCoding73.4% ±2.052 / 87
ScreenSpot ProMultimodal68.2%13 / 20
Humanity's Last ExamKnowledge28.8%51 / 77
Vibe Code Bench v1.1Coding25.6% ±3.984 / 123
Claw-EvalAgentic58.8%29 / 42
LiveBench LanguageKnowledge75.0%47 / 66
LiveBench Data AnalysisReasoning69.9%57 / 80
Terminal-Bench 2.1 · Vals AIAgentic53.2% ±1.656 / 75
FrontierMath-Tiers-1-3-v2-PrivateMath32.3% ±2.868 / 90
Mystery Game PuzzlesReasoning12.0% ±3.361 / 80
ToolathlonAgentic39.8%17 / 22
NL2RepoCoding37.9%28 / 36
Terminal-Bench HardAgentic43.9%18 / 23
Chess PuzzlesReasoning16.0% ±3.771 / 90
Code MigrationCoding11.1% ±1.364 / 77
LiveBench MathematicsMath83.7%56 / 66
LiveBench ReasoningReasoning75.8%68 / 80
MCP AtlasAgentic48.2%45 / 52
LiveBench Instruction FollowingInstruction58.3%58 / 66
SWE-Bench verified · Epoch AICoding57.9% ±2.229 / 32
LiveBench Agentic CodingAgentic41.4%60 / 66
FrontierMath-Tier-4-2025-07-01-PublicMath0.0%3 / 36
MortgageTaxFinance68.0% ±0.920 / 97
TaxEval v2Finance74.7% ±0.931 / 143
LegalBenchLegal84.2% ±0.437 / 149
FrontierMath-2025-02-28-PublicMath50.0% ±16.716 / 64
Finance Agent v1.1Finance54.6% ±2.915 / 51
CorpFin v2Finance61.9% ±1.049 / 131
ProgramBenchCoding0.0%34 / 65
MedScribeHealthcare77.0% ±1.973 / 109
MedCodeHealthcare36.9% ±2.076 / 106
Finance Agent (v2)Finance40.8% ±0.163 / 79
Legal Research BenchLegal14.9% ±2.575 / 90
EMBFinance32.9% ±2.863 / 75
Tax Agent BenchFinance7.7% ±1.872 / 82
CaseLaw v2Legal51.4% ±1.248 / 53
VideoMMMUMultimodal84.0%—
MMAnswerBenchMath83.8%—
MCP-TasksAgentic74.1%—
AI-NeedleReasoning68.3%—
NOVA-63Multilingual57.9%—
DeepPlanningAgentic41.5%—
ResearchClawBenchAgentic18.0%—

Sources

BenchLMCC BY-NC 4.0OpenRouterTerms restrict reuseVals AINo licence statedEpoch AICC BYLiveBenchNo licence statedmodels.devMIT

History

5 releases
Index55.5 → 54.6
23 Oct 202612 Nov 2026
Output / 1M$1.95 → $1.95
23 Oct 202612 Nov 2026

Questions

How much does Qwen3.6 Plus cost?
$0.325 per 1M input tokens, $1.95 per 1M output tokens.
How good is Qwen3.6 Plus?
Its index is 54.6 out of 100. That is place 107 of 283 ranked models.
What is Qwen3.6 Plus best at?
Knowledge: 54.1 out of 100, place 102 of 268.
How fast is Qwen3.6 Plus?
44 output tokens per second, 1.1 s to the first token.
What is the context window of Qwen3.6 Plus?
1M tokens.
Is Qwen3.6 Plus open source?
No. Its weights are not published.
When was Qwen3.6 Plus released?
April 2026.

Badge

Qwen3.6 Plus on Inference360: place 107 of 283, index 54.6
HTML
<a href="https://inference360.ai/models/qwen3-6-plus"><img src="https://inference360.ai/badges/qwen3-6-plus.svg" alt="Qwen3.6 Plus on Inference360: place 107 of 283, index 54.6" height="20"></a>
Markdown
[![Qwen3.6 Plus on Inference360: place 107 of 283, index 54.6](https://inference360.ai/badges/qwen3-6-plus.svg)](https://inference360.ai/models/qwen3-6-plus)

Compare with

GPT-5.4 mini54.5Inkling54.8MiMo-V2.554.8Qwen3.5 397B53.8Qwen3.6-35B-A3B50.6

Lower price

dots3-note Preview63.4 · FreeApodex 1.1 Mini55.7 · FreeGLM-5.3-Flash63.1 · $0.078/1M

More from Alibaba

All models
Qwen3.6-35B-A3B50.6Qwen 3.6 Max (preview)57.2Qwen3.6-27B52.1Qwen3.5 9BUnrankedQwen3.6 FlashUnranked