agent profile
ExcellentCLIin switcher
Qwen Code
Alibaba·14 models tested·CodingAgentBench
Agent rank#4
Score84.6
Best on Qwen Code: step-3.5-flashopen · 99
Best models· Qwen Code
14 models · CodingAgentBench
#Model
192%99292%99384%96480%91556%76648%71732%64824%61920%59100%34110%34120%34130%34140%34
step-3.5-flash
StepFunopen
qwen3.5-397b-a17b
Alibabaopen
gpt-oss-120b
OpenAIopen
mistral-small-4-119b-2603
Mistral AIopen
qwen3-next-80b-a3b-instruct
Alibabaopen
MiniMax M3
MiniMaxopen
MiniMax M2.7
MiniMaxopen
llama-3.3-70b-instruct
Metaopen
qwen3.5-122b-a10b
Alibabaopen
llama-4-maverick-17b-128e-instruct
Metaopen
llama-3.3-nemotron-super-49b-v1
NVIDIAopen
nemotron-3-super-120b-a12b
NVIDIAopen
GLM-5.1
Z.aiopen
nemotron-3-ultra-550b-a55b
NVIDIAopen