ResearchBenchmarksTool Use on APT-BenchFollow65.8AccuracyQwen356.33658.79361.2563.707Dec 31, 2025Evaluation ResultsMethodMethodLinksAccuracyQwen3Parameters=4B, Variant...Parameters=4B, Variant=Base2025.1265.8Youtu-LLMParameters=2B, Variant...Parameters=2B, Variant=Base2025.1264.2Llama3.1Parameters=8B, Variant...Parameters=8B, Variant=Base2025.1264.1Gemma3Parameters=4B, Variant...Parameters=4B, Variant=Base2025.1261.7SmolLM3Parameters=3B, Variant...Parameters=3B, Variant=Base2025.1259.1Qwen3Parameters=1.7B, Varia...Parameters=1.7B, Variant=Base2025.1256.7