Inference
EN
Alibaba

Qwen3.6-27B

Qwen3.6-27B is a reasoning model from Alibaba. 47 benchmarks count toward its index, in 8 categories.

  • Open weights
  • Reasoning
  • Input: text · image · video
  • Released 2026-04
RankedEnough results for an index.
IndexAll categories, out of 100.52.1±2.4Median 51.0
CoverageIndex weight that has results.100%
SpeedOutput tokens per second.60/sMedian 67/s
First tokenSeconds to the first output token.0.6 sMedian 1.3 s
Input / 1MUS dollars per 1M input tokens.$0.32Median $0.60
Output / 1MUS dollars per 1M output tokens.$3.20Median $2.55
ContextMaximum tokens in one request.262KMedian 262K
EloLMArena rating and rank.—Median 1422

Details

From OpenRouter
API IDThe model ID on OpenRouter.
qwen/qwen3.6-27b
Max outputMaximum output tokens per request.
82K
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.
Hugging Face (opens in a new tab) ↗
RetiresDate OpenRouter removes the model.
—

Where to run

APIs that serve this model.
8 providers
ProviderTypeMaker, cloud, host or router.InputOutputContext
STACKITDeprecatedCloud$0.53$0.76262K
ChutesHost$0.25$2.00262K
GroqDeprecatedHost$0.60$3.00131K
OVHcloudCloud$0.47$3.19262K
SiliconFlowHost$0.30$3.20262K
Deep InfraHost$0.32$3.20262K
OpenRouterRouter$0.32$3.20262K
Novita AIHost$0.60$3.60262K

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

Run locally

27.78B parameters
16-bit
59.3 GB
8-bit
32.0 GB
4-bit
18.7 GB
Weights
Hugging Face (opens in a new tab) ↗

Memory need with a short context. The 4-bit copy fits in 24 GB. Compare local models

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.9 counted56.791 / 218
CodingCode writing and repair.7 counted52.8121 / 209
ReasoningLogic problems and puzzles.4 counted49.0122 / 227
MultimodalTasks with images and text.9 counted53.462 / 84
KnowledgeFacts and expert knowledge.8 counted48.2147 / 268
MultilingualTasks in many languages.1 counted60.9—
InstructionTasks with strict prompt rules.2 counted43.241 / 63
MathMath problems.7 counted51.082 / 168

Results

47 counted
BenchmarkThe benchmark name.ResultThe score from the source.PlacePlace among results on this benchmark.
EuroEval PolishMultilingual65.3% ±0.56 / 209
CountBenchMultimodal97.8%1 / 10
RefCOCO averageMultimodal92.5%1 / 10
Massive Multitask Language Understanding ProfessionalKnowledge86.2%12 / 75
τ²-Bench Tool-Agent-User EvaluationAgentic94.2%26 / 149
Claw-EvalAgenticAbove fitFar above this model's index.72.4%8 / 42
Massive Multi-discipline Multimodal UnderstandingMultimodal82.9%7 / 28
Graduate-Level Google-Proof Q&AKnowledge87.8%40 / 141
Software Engineering Benchmark VerifiedCoding77.2%36 / 114
Gert Labs Composite Game BenchmarkAgentic54.8%17 / 52
GPQA diamond · Epoch AIKnowledge84.8% ±2.672 / 217
Artificial Analysis Long Context ReasoningReasoning77.3%77 / 217
V*Multimodal94.7%5 / 14
Artificial Analysis IFBenchInstruction67.6%53 / 146
Terminal-Bench 2.0 · Vals AIAgentic44.9% ±5.324 / 66
MMLU-ReduxKnowledge93.5%7 / 19
RealWorldQAMultimodal84.1%7 / 18
LiveCodeBench v6Coding83.9%24 / 59
C-EvalKnowledge91.4%5 / 12
Artificial Analysis GPQA DiamondKnowledge84.2%83 / 199
Terminal-Bench 2.0 · BenchLMAgentic59.3%18 / 41
Harvard-MIT Mathematics Tournament February 2025Math93.8%7 / 15
Artificial Analysis Humanity's Last ExamKnowledge23.1%104 / 222
Artificial Analysis MMMU-ProMultimodal74.6%57 / 113
OTIS Mock AIME 2024-2025Math66.7% ±7.1106 / 200
Critical Physics TasksReasoning1.1%118 / 214
GDPval-AA normalizedAgentic24.3%75 / 133
AIME 2026Math94.1%20 / 34
SuperGPQA: Scaling LLM Evaluation Across 285 Graduate DisciplinesKnowledge66.0%14 / 22
Massive Multi-discipline Multimodal Understanding ProMultimodal75.8%34 / 53
Harvard-MIT Mathematics Tournament November 2025Math90.7%8 / 12
Harvard-MIT Mathematics Tournament February 2026Math84.3%18 / 27
LiveBench Data AnalysisReasoning70.4%55 / 80
Artificial Analysis SciCodeCoding42.8%79 / 114
QwenClawBenchAgentic53.4%7 / 10
SWE-benchCoding70.0% ±2.161 / 87
ERQAMultimodal62.5%13 / 18
FrontierMath-Tiers-1-3-v2-PrivateMath34.0% ±2.865 / 90
SWE-bench ProCoding53.5%60 / 83
CharXiv ReasoningMultimodal78.4%38 / 49
Humanity's Last ExamKnowledge24.0%60 / 77
NL2RepoCoding36.2%29 / 36
Artificial Analysis Omniscience AccuracyKnowledge19.6%134 / 164
LiveBench CodingCoding71.8%55 / 66
SimpleVQAMultimodal56.1%15 / 18
Vibe Code Bench v1.1Coding11.9% ±3.4103 / 123
Mystery Game PuzzlesReasoning7.0% ±2.669 / 80
LiveBench MathematicsMath79.9%58 / 66
AndroidWorldAgentic70.3%9 / 10
LiveBench LanguageKnowledge63.3%62 / 66
LiveBench ReasoningReasoning70.3%76 / 80
LiveBench Agentic CodingAgentic39.3%63 / 66
LiveBench Instruction FollowingInstruction53.2%65 / 66
MortgageTaxFinance68.3% ±0.917 / 97
CorpFin v2Finance62.3% ±1.047 / 131
TaxEval v2Finance71.3% ±0.979 / 143
CaseLaw v2Legal53.2% ±2.240 / 53
Video-MME with subtitleMultimodal87.7%—
MLVU mean averageMultimodal86.6%—
DynaMathMultimodal85.6%—
VideoMMMUMultimodal84.4%—
MStarMultimodal81.4%—
CC-OCRMultimodal81.2%—
MMAnswerBenchMath80.8%—

Sources

BenchLMCC BY-NC 4.0OpenRouterTerms restrict reuseEpoch AICC BYLiveBenchNo licence statedVals AINo licence statedEuroEvalMITmodels.devMIT

History

5 releases
Index52.9 → 52.1
23 Oct 202612 Nov 2026
Output / 1M$2.00 → $3.20
23 Oct 202612 Nov 2026

Questions

How much does Qwen3.6-27B cost?
$0.32 per 1M input tokens, $3.20 per 1M output tokens.
How good is Qwen3.6-27B?
Its index is 52.1 out of 100. That is place 129 of 283 ranked models.
What is Qwen3.6-27B best at?
Agentic: 56.7 out of 100, place 91 of 218.
How fast is Qwen3.6-27B?
60 output tokens per second, 0.6 s to the first token.
What is the context window of Qwen3.6-27B?
262K tokens.
Is Qwen3.6-27B open source?
Yes. Its weights are published.
When was Qwen3.6-27B released?
April 2026.

Badge

Qwen3.6-27B on Inference360: place 129 of 283, index 52.1
HTML
<a href="https://inference360.ai/models/qwen3-6-27b"><img src="https://inference360.ai/badges/qwen3-6-27b.svg" alt="Qwen3.6-27B on Inference360: place 129 of 283, index 52.1" height="20"></a>
Markdown
[![Qwen3.6-27B on Inference360: place 129 of 283, index 52.1](https://inference360.ai/badges/qwen3-6-27b.svg)](https://inference360.ai/models/qwen3-6-27b)

Compare with

Solar Pro 452.1Gemini 3.5 Flash-Lite52.0Ternary Bonsai 2 27B52.2Claude Opus 4.552.0Qwen 3.6 Max (preview)57.2

Lower price

dots3-note Preview63.4 · FreeApodex 1.1 Mini55.7 · FreeNemotron 3 Ultra50.8 · Free

More from Alibaba

All models
Qwen 3.6 Max (preview)57.2Qwen3.6-35B-A3B50.6Qwen3.6 FlashUnrankedQwen3.6 Plus54.6Qwen3.7 Max60.5