Codesota · Papers2026-02-16
Paper

Qwen3.5: Towards Native Multimodal Agents

Paper ↗Code ↗
§ 01 · Benchmark results

42 results reproduced from this paper.

View:
Sorted instantly in-page
Results
41
SOTA rows
0
Models
4
Datasets
0
#ModelVendorBenchmarkValueSOTADateSource
01Qwen3.5-397B-A17BAlibabaOmniDocBench90.8%source ↗
02Qwen3.5-397B-A17BAlibabaGPQA Diamond88.4%source ↗
03Qwen3.5-397B-A17BAlibabaVideo-MME87.5%source ↗
04Qwen3.5-397B-A17BAlibabaTau2-Bench86.7%source ↗
05Qwen3.5-122B-A10BAlibabaGPQA Diamond86.6%source ↗
06Qwen3.5-27BAlibabaMMLU-Pro86.1%source ↗
07Qwen3.5-27BAlibabaGPQA Diamond85.5%source ↗
08Qwen3.5-35B-A3BAlibabaMMLU-Pro85.3%source ↗
09Qwen3.5-122B-A10BAlibabaRealWorldQA85.1%source ↗
10Qwen3.5-397B-A17BAlibabaMMMU85.0%source ↗
11Qwen3.5-35B-A3BAlibabaGPQA Diamond84.2%source ↗
12Qwen3.5-35B-A3BAlibabaRealWorldQA84.1%source ↗
13Qwen3.5-122B-A10BAlibabaMMMU83.9%source ↗
14Qwen3.5-397B-A17BAlibabaRealWorldQA83.9%source ↗
15Qwen3.5-397B-A17BAlibabaMMStar83.8%source ↗
16Qwen3.5-27BAlibabaRealWorldQA83.7%source ↗
17Qwen3.5-122B-A10BAlibabaMMStar82.9%source ↗
18Qwen3.5-27BAlibabaMMMU82.3%source ↗
19Qwen3.5-35B-A3BAlibabaMMStar81.9%source ↗
20Qwen3.5-35B-A3BAlibabaMMMU81.4%source ↗
21Qwen3.5-35B-A3BAlibabaTau2-Bench81.2%source ↗
22Qwen3.5-27BAlibabaMMStar81.0%source ↗
23Qwen3.5-122B-A10BAlibabaTau2-Bench79.5%source ↗
24Qwen3.5-397B-A17BAlibabaMMMU-Pro79.0%source ↗
25Qwen3.5-27BAlibabaTau2-Bench79.0%source ↗
26Qwen3.5-397B-A17BAlibabaMVBench77.6%source ↗
27Qwen3.5-122B-A10BAlibabaMMMU-Pro76.9%source ↗
28Qwen3.5-122B-A10BAlibabaMVBench76.6%source ↗
29Qwen3.5-397B-A17BAlibabaSWE-Bench Verified76.4%source ↗
30Qwen3.5-35B-A3BAlibabaMMMU-Pro75.1%source ↗
31Qwen3.5-27BAlibabaMMMU-Pro75.0%source ↗
32Qwen3.5-35B-A3BAlibabaMVBench74.8%source ↗
33Qwen3.5-27BAlibabaMVBench74.6%source ↗
34Qwen3.5-27BAlibabaSWE-Bench Verified72.4%source ↗
35Qwen3.5-122B-A10BAlibabaSWE-Bench Verified72.0%source ↗
36Qwen3.5-35B-A3BAlibabaSWE-Bench Verified69.2%source ↗
37Qwen3.5-397B-A17BAlibabaBrowseComp69.0%source ↗
38Qwen3.5-122B-A10BAlibabaBrowseComp63.8%source ↗
39Qwen3.5-27BAlibabaBrowseComp61.0%source ↗
40Qwen3.5-35B-A3BAlibabaBrowseComp61.0%source ↗
41Qwen3.5-397B-A17BAlibabaHLE28.7%source ↗
Benchmark trail
§ 02 · Models

4 models from this paper.

evaluates
Qwen3.5-397B-A17B
Alibaba
evaluates
Qwen3.5-35B-A3B
Alibaba
evaluates
Qwen3.5-27B
Alibaba
evaluates
Qwen3.5-122B-A10B
Alibaba
Read next

Three places to go from here.

Index
All papers
All tracked papers in the registry, with benchmark result, model, and leaderboard linkage where available.
Replacement
Papers with Code is dead — alternatives
What replaced PWC for each use case: LLMs, OCR, speech, vision, robotics.
Top hub
LLM benchmarks
Every frontier LLM benchmark, scored.