Models · Qwen (Alibaba) · Out sinceReleased 22 Apr 2026

Qwen3.6 27B

Available on OpenRouterOn OpenRouterAvailable nowGenerally availableCan run on your own serversOpen weights

Qwen3.6 27B is made by Qwen (Alibaba). We don't have enough test results yet to rank it. It's cheap to use. Its makers have published it, so you can run it on your own servers.

Open-weight dense 27B; thinking and non-thinking in one checkpoint. Superseded by Qwen3.8-27B.

Writing & creativity: —
Research & analysis: —
Coding: —
Price: Cheap$0.32 / $3.20
Price per 1M (blended)Blended / 1M: $1.04
MemoryContext: 262K
Longest answerMax output: —
Test resultsResults: 18 (2 independent2 indep.)
Out sinceReleased: 22 Apr 2026
Made byVendor: Qwen (Alibaba)
UnderstandsInputs: text, image, video

Scores are out of 100 for each area, compared with every model we track; “#” is its rank. Context is how much text it can read at once (1M is roughly 700,000 words). Blended price mixes the cost of what you send and what it writes back.

Thinking levels

Reasoning settings

No reasoningDefault

Links

Using it through OpenRouter

On OpenRouter

OpenRouter is a service that gives access to many AI models in one place. This model is listed there as qwen/qwen3.6-27b.

Route it as qwen/qwen3.6-27b at $0.32 in / $3.20 out per 1M tokens, 262K context. Listed since 27 Apr 2026.

frequency_penaltyinclude_reasoninglogit_biaslogprobsmax_tokensmin_ppresence_penaltyreasoningrepetition_penaltyresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_ktop_logprobstop_p

Run it yourself

Self-hosting

On your own servers,Licence, sizeunder your own control.and hardware.

Compare all open models →All open-weights models →

Details comingselfHost data pending

This model can be downloaded and run on your own servers. We're collecting its licence, size and hardware needs now.Open-weights model; licence, parameter count, hardware tier and base-model availability will appear after the next data run.

Every test result

Every result

All the numbers,with where they came from.

Every result we've found for this model, with who measured it and a link to where we read it.

All raw rows (vendor and independent kept separate), with setting label, source, date, harness and notes.

All results for this model
						NotesNotes
AIME 2026	94.1%	Defaultthinking	Maker's own figureVendor-reportedQwen (Alibaba) ↗	22 Apr 2026	—
DeepSWE	13.3%	Defaultsetting not stated	Maker's own figureVendor-reportedQwen (Alibaba) ↗	14 Aug 2026	Claude Code	Qwen3.6-27B column in the Qwen3.8-27B model card
EuroEval Swedish (generative)	1.5	DefaultQwen/Qwen3.6-27B-FP8 (val)	Independent testIndependentEuroEval (Alexandra Institute) ↗	29 Sep 2026	—	EuroEval rank tier 3; ±0.19; lower is better; task scores (first metric): SweDN summarisation 37.29 ± 0.24, Skolprov 81.78 ± 2.64, Swedish facts 35.07 ± 2.06, ScaLA-sv 63.98 ± 1.94
GPQA Diamond	87.8%	Defaultthinking	Maker's own figureVendor-reportedQwen (Alibaba) ↗	22 Apr 2026	—
HMMT February 2026	84.3%	Defaultthinking	Maker's own figureVendor-reportedQwen (Alibaba) ↗	22 Apr 2026	—	HMMT Feb 26
Humanity's Last Exam	24.0%	Defaultthinking	Maker's own figureVendor-reportedQwen (Alibaba) ↗	22 Apr 2026	—
IMO-AnswerBench	80.8%	Defaultthinking	Maker's own figureVendor-reportedQwen (Alibaba) ↗	22 Apr 2026	—
LiveCodeBench	83.9%	Defaultthinking	Maker's own figureVendor-reportedQwen (Alibaba) ↗	22 Apr 2026	—	LiveCodeBench v6
MMLU-Pro	86.2%	Defaultthinking	Maker's own figureVendor-reportedQwen (Alibaba) ↗	22 Apr 2026	—
MMMU-Pro (no tools)	75.8%	Defaultthinking	Maker's own figureVendor-reportedQwen (Alibaba) ↗	22 Apr 2026	—
NL2Repo-Bench	36.2%	Defaultthinking	Maker's own figureVendor-reportedQwen (Alibaba) ↗	22 Apr 2026	Claude Code
SkillsBench	48.2%	Defaultthinking	Maker's own figureVendor-reportedQwen (Alibaba) ↗	22 Apr 2026	OpenCode
SWE-bench Multilingual	71.3%	Defaultthinking	Maker's own figureVendor-reportedQwen (Alibaba) ↗	22 Apr 2026	Internal scaffold (bash + file-edit)
SWE-Bench Pro (public, v1)	53.5%	Defaultthinking	Maker's own figureVendor-reportedQwen (Alibaba) ↗	22 Apr 2026	Internal scaffold (bash + file-edit)	Qwen-corrected SWE-bench Pro
SWE-bench Verified	77.2%	Defaultthinking	Maker's own figureVendor-reportedQwen (Alibaba) ↗	22 Apr 2026	Internal scaffold (bash + file-edit)
SWE-rebench	31.2%	DefaultQwen3.6-27B	Independent testIndependentSWE-rebench (Nebius) ↗	1 Oct 2026	SWE-rebench standard scaffold	time window 2026-05-15..2026-07-01 (111 problems, 65 repos); ±1.68; pass@5 57.7%; $0.62/problem
Terminal-Bench 2.0	59.3%	Defaultthinking	Maker's own figureVendor-reportedQwen (Alibaba) ↗	22 Apr 2026	Terminus 2 (Harbor)	avg of 5 runs
Terminal-Bench 2.1	63.4%	Defaultsetting not stated	Maker's own figureVendor-reportedQwen (Alibaba) ↗	14 Aug 2026	Terminus 2	Qwen3.6-27B column in the Qwen3.8-27B model card