Model

Qwen3.6-Plus

CLOSED

Alibaba · Qwen3.6 family · released Apr 2, 2026

Flagship hosted model of the Qwen3.6 generation ('Towards Real World Agents'); massive agentic-coding upgrade over Qwen3.5-Plus per the launch blog, with a 1M-token default context and improved multimodal perception/reasoning. This is the model matching the leads' 'Qwen 3.6 Plus' entry; a second, separately-dated lead for the same product name was found to be a duplicate (see excluded).

ReasoningCodingVisionFunction callingTool useAgentic

2290.1

Elo · rank #50

Parameters

Undisclosed

Active params

Undisclosed

Context

1M tokens

Architecture

Hosted product tier of the Qwen3.6 hybrid-attention line (Gated DeltaNet + Gated Attention, Qwen3-Next lineage); exact parameter/expert/MoE configuration for the hosted Plus checkpoint undisclosed

License

Proprietary

Languages

—

API price (in/out)

$0.5 / $3

Modalities

text · vision · video

Benchmark results

Bar shows position within the tracked field; marker = field best

Agents' Last ExamAgents24.3%#20

best: GPT-6 Astra · 59.3%

AI2DVision94.4%#9

best: Molmo 72B · 96.3%

AIMEMath95.3%#15

best: GPT-5.2 · 100.0%

C-EvalKnowledge93.3%#1

best: this model · 93.3%

CharXivVision81.5%#11

best: Qwen3.8-Flash-Next · 90.6%

GDPval-AAAgents1135#40

best: Claude Fable 5 · 1932

GPQA DiamondReasoning88.2%#41

best: GPT-6 Astra · 96.0%

Humanity's Last ExamReasoning25.7%#58

best: Claude Opus 5 · 64.7%

IFBenchReasoning75.2%#20

best: MiniMax M3 · 83.0%

IFEvalReasoning94.3%#5

best: Gemma 4 26B A4B · 98.5%

LiveCodeBenchCoding87.1%#18

best: DeepSeek-V4-Pro (Think Max) · 93.5%

MathVisionVision88.0%#6

best: Seed 2.1 Pro · 92.6%

MMLU-ProKnowledge88.5%#6

best: Claude Fable 5 · 91.5%

MMLU-ReduxKnowledge94.5%#3

best: Qwen3.7-Max · 95.0%

MMMLUKnowledge89.5%#8

best: Gemini 3.1 Pro · 92.6%

MMMU-ProVision78.8%#16

best: Claude Opus 4.7 · 85.5%

MMMUVision86.0%#6

best: Claude Fable 5 · 89.3%

OSWorld-VerifiedAgents62.5%#23

best: Claude Fable 5 · 85.0%

RefCOCOVision93.5%#3

best: Qwen3.5-Omni-Plus · 95.0%

SuperGPQAReasoning71.6%#2

best: Qwen3.7-Max · 73.6%

SWE-bench ProCoding56.6%#28

best: Claude Fable 5.1 · 81.2%

SWE-bench VerifiedCoding78.8%#25

best: Claude Opus 5 · 96.0%

τ²-Bench TelecomAgents97.7%#12

best: Claude Opus 4.6 · 99.3%

Terminal-Bench 2.0Coding61.6%#41

best: Gemini 3.8 Flash · 89.4%

Video-MMEVision87.8%#3

best: Seed 2.1 Pro · 89.2%

Run it locally

Closed weights — available via API only. No local deployment.

Input / M tok

$0.5

Output / M tok

Qwen3.6 family

Elo progression across releases

Qwen3.6-PlusApr 20262290.1 Qwen3.6-35B-A3BApr 20261973.2 Qwen3.6-27BApr 20262027.4

Resources

DOCShttps://qwen.ai/blog?id=qwen3.6 APIAvailable via Alibaba API

API price $0.5/$3 · each benchmark row carries its own source badge (see methodology)