Qwen 3.5 27B
Qwen 3.5 27BRankingin thinking mode?
Alternatives?
CheaperMiMo v2.6 ProunstatedOpenwins 57%-34%$0.43 / $0.87Memory (4-bit): 17 GB → 615 GB+4
BetterGPT-5.6 Terraxhighwins 55%+445%$0.63Prompts with >272K input tokensOpenQwen3.8 27Bunstatedwins 53%+16%$0.42 / $2.55max_tokens range · QwenCloud+4
CheaperGLM 5.3 FlashunstatedOpenwins 57%-71%$0.15 / $0.50Memory (4-bit): 17 GB → 193 GB+5CheaperMiMo v2.6 FlashunstatedOpenwins 55%-79%$0.14 / $0.28Memory (4-bit): 17 GB → 186 GB+4MiMo v2.6 ProunstatedOpenwins 57%-34%$0.43 / $0.87Memory (4-bit): 17 GB → 615 GB+4GLM 5.3 FlashunstatedOpenwins 57%-71%$0.15 / $0.50Memory (4-bit): 17 GB → 193 GB+5MiMo v2.6 FlashunstatedOpenwins 55%-79%$0.14 / $0.28Memory (4-bit): 17 GB → 186 GB+4Qwen3.7 Plusunstatedwins 55%-32%$0.32 / $1.28MiMo v2.5 ProunstatedOpenwins 54%-34%$0.43 / $0.87Memory (4-bit): 17 GB → 614 GB+1GPT-5.6 Terraxhighwins 55%+445%$0.63Prompts with >272K input tokensGPT-6.1 Solmaxwins 52%+385%$0.72Remove unsupported parameters+5
Muse Spark 1.2xhighwins 54%+142%$0.97Only auto tool_choice is supported+1GPT-5.6 Solxhighwins 57%+870%$1.18
Gemini 3.8 Flashhighwins 60%+82%*$1.24Sampling parameters are ignored+1Qwen3.8 27BunstatedOpenwins 53%+16%$0.42 / $2.55max_tokens range · QwenCloud+4Gemma 4 26B A4BunstatedOpenwins 53%-68%$0.15 / $0.60Context: 262K → 256KGemma 4 31BunstatedOpensame-81%$0.10 / $0.34Context: 262K → 256K+1Qwen 3.5 35B A3BunstatedOpenwins 47%-17%$0.25 / $2Memory (4-bit): 17 GB → 22 GBQwen3 30B A3B Instruct 2507unstatedOpenwins 44%-83%$0.09 / $0.30Memory (4-bit): 17 GB → 18 GB+1Qwen 3.7 Max Previewwins 57%—$2.50 / $7.50Context: not verified+1Qwen3.7 Plusunstatedwins 55%—$0.32 / $1.28Qwen3.8 27BunstatedOpenwins 53%—$0.42 / $2.55max_tokens range · QwenCloud+4Qwen3.5 397B A17BunstatedOpenwins 53%—$0.60 / $3.60Memory (4-bit): 17 GB → 242 GBQwen3.6 Max Previewunstatedsame—$1.30 / $7.80No image input+1Qwen3 235B A22B Thinking 2507unstatedOpensame—$0.23 / $2.30Memory (4-bit): 17 GB → 141 GB+2Qwen3 Max (Preview)unstatedsame—$0.87 / $3.48No image inputQwen3.6 Plusunstatedsame—$0.50 / $3All Alibaba (Qwen) models →Run it yourself
estimateHistory
3 versionsBenchmarks
all categoriesFAQ
How much does Qwen 3.5 27B cost?›
$0.30 per 1M input tokens and $2.40 per 1M output tokens; cheapest via lab-qwen at $0.086 / $0.69.
Is Qwen 3.5 27B free to use?›
There is no free API; the cheapest paid price is $0.086 per 1M input tokens. The weights are open, so you can run it yourself.
How good is Qwen 3.5 27B, and what is it best at?›
#102 of 334 in Chat & Knowledge, #79 of 97 in Coding, #26 of 242 in Agents & Tool Use, #97 of 195 in Writing & Language, #49 of 124 in Vision & Documents. Strongest at: Tool calling, Science, Instructions.
Is Qwen 3.5 27B better than Qwen3 Max (Preview)?›
On Arena Text · Chinese, Qwen 3.5 27B scores 1,478 vs 1,478.2 — about the same.
How can I use Qwen 3.5 27B?›
Through the API of Alibaba, AtlasCloud, DeepInfra, Novita, SiliconFlow.
What is the context window of Qwen 3.5 27B?›
262K tokens, with up to 66K output tokens.
When was Qwen 3.5 27B released? Is there a newer version?›
Released 2026-02-23. It has been superseded by Qwen3.6 27B.
Is Qwen 3.5 27B open source?›
Yes, the weights are open (Apache-2.0 licence).
Can I run Qwen 3.5 27B locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can Qwen 3.5 27B read images?›
It accepts text and images, video as input.
What is Qwen 3.5 27B?›
Qwen 3.5 27B is an open-weights language model from Alibaba (Qwen), released Feb 2026, the successor to Qwen3 VL 32B (Reasoning).
What are the best alternatives to Qwen 3.5 27B?›
MiMo v2.6 Pro (cheaper), GPT-5.6 Terra (better), Qwen3.8 27B (open weights).