agent profile
SolidAgentin switcher
mini-SWE-agent
Princeton·11 models tested·ixio runs + NL2Repo + Terminal-Bench 2.1
Agent rank#5
Score80.0
Best on mini-SWE-agent: Claude Fable 5 · 100
Best models· mini-SWE-agent
11 models · NL2Repo + Terminal-Bench 2.1
#Model
1——1002——1003——1004100%—1005—76.2%59655.4%—507——428——319——2310——811——8
Claude Fable 5
Anthropic
Claude Opus 4.8
Anthropic
DeepSeek-V4
DeepSeekopen
MiniMax M3
MiniMaxopen
Muse Spark 1.1
Metaopen
Kimi K2.7 Code
Moonshot AIopen
GLM-5.2
Z.aiopen
MiniMax M2.5
MiniMaxopen
Llama 3.3 70B
Metaopen
Kimi K2.6
Moonshot AIopen
gpt-oss-120b
OpenAIopen