List your company

Agentic Instruction Following on TaskCraft

79.33Performance ScoreMemEvolve + Flash-Searcher
57.146862.905968.66574.4241Dec 21, 2025

Evaluation results

MethodLinks
MemEvolve + Flash-SearcherModel Family=GPT-5-mini, pass@k=32025.1279.33
MemEvolve + SmolAgentModel Family=GPT-5-mini, pass@k=32025.1277
Agent KBModel Family=GPT-4.1, pass@k=32025.1275.33
MemEvolve + SmolAgentModel Family=GPT-5-mini, pass@k=22025.1275
MemEvolve + Flash-SearcherModel Family=GPT-5-mini, pass@k=22025.1275
Agent KBModel Family=GPT-4.1, pass@k=22025.1272.67
MemEvolve + Flash-SearcherModel Family=DeepSeek V3.2, pass@k=12025.1272.67
MemEvolve + Flash-SearcherModel Family=GPT-5-mini, pass@k=12025.1272
Flash-SearcherModel Family=GPT-5-mini, pass@k=12025.1269.67
Flash-SearcherModel Family=DeepSeek V3.2, pass@k=12025.1269.33
MemEvolve + Flash-SearcherModel Family=Kimi K2, pass@k=12025.1268
MemEvolve + SmolAgentModel Family=GPT-5-mini, pass@k=12025.1267.67
Cognitive Kernel-ProModel Family=Claude-3.7 etc., pass@k=12025.1266
SmolagentsModel Family=GPT-5-mini2025.1264
Agent KBModel Family=GPT-4.1, pass@k=12025.1261.67
OWL WorkforceModel Family=GPT-4o+o3-mini, pass@k=32025.1258.33
Flash-SearcherModel Family=Kimi K2, pass@k=12025.1258