Our new X account is live! Follow @wizwand_team for updates

Agentic Coding on SWE-bench Verified

77.2Percentage Resolved

Claude-Sonnet-4.5

Updated 4d ago

Evaluation Results

Method	Links
Claude-Sonnet-4.5 2025.12		77.2
GPT-5 2025.12		74.9
Kimi-K2-thinking 2025.12		71.3
DeepSeek-V3.1-Nex-N1 2025.12		70.6
Minimax-M2 2025.12		69.4
GLM-4.6 2025.12		68
DeepSeek-V3.1 2025.12		66
Gemini-2.5-pro 2025.12		59.6
Qwen3-32B-Nex-N1 2025.12		50.5
Qwen3-30B-A3B-Nex-N1 2025.12		29.7
InternLM3-8B-Nex-N1 2025.12		20.3
ALIVE-Oracle 2026.02		17.6
ALIVE-Self 2026.02		17.2
GRPO (Scalar Reward) 2026.02		14.8
FCP (Verbal Only) 2026.02		14
SFT 2026.02		13.6
Qwen3-32B 2025.12		12.9
Base Model 2026.02		11.8
Qwen3-30B-A3B 2025.12		9.6