Behavior Tree Generation on LLM-based BT Generation Benchmark benchmark leaderboard · SOTA2 Research

Behavior Tree Generation on LLM-based BT Generation Benchmark

5.1Execution Time (Easy)BTGenBot-2

Evaluation Results

Method	Links
BTGenBot-22026.02		5.1	—	—	—	—	7.5	—	—	11	84.61	100	96.15	94.23
BTGenBot-2-ER2026.02		5.8	—	—	—	—	8.2	—	—	11.7	90.38	100	100	94.23
BTGenBot-2-ER2026.02		8.4	—	—	—	—	9.6	—	—	13.1	98.07	100	100	98.07
BTGenBot-22026.02		8.4	—	—	—	—	8.9	—	—	12.6	92.38	100	98.07	94.23
GPT-5 Instant2026.02		12.8	—	—	—	—	15	—	—	15.2	65.38	88.46	82.69	80.76
GPT-5 Instant2026.02		15.3	—	—	—	—	16.2	—	—	18.8	78.84	96.15	86.53	84.61
Claude Opus 4.12026.02		23.1	—	—	—	—	27.4	—	—	29.8	65.38	90.38	82.69	80.76
Claude Opus 4.1-ER2026.02		23.9	—	—	—	—	28.1	—	—	30.4	76.92	92.3	90.38	88.46
Claude Opus 4.12026.02		25	—	—	—	—	29.4	—	—	33.6	84.61	96.15	90.38	88.46
Claude Opus 4.1-ER2026.02		25.5	—	—	—	—	29.8	—	—	34.3	88.46	100	96.15	94.23
GPT-5 Thinking2026.02		34.2	—	—	—	—	36.8	—	—	39.7	71.15	92.3	86.53	86.53
GPT-5 Thinking2026.02		35.8	—	—	—	—	36.8	—	—	42.2	84.61	100	92.3	88.46
BTGenBot2026.02		85	—	—	—	—	98	—	—	178	57.69	63.46	78.84	46.15
BTGenBot2026.02		101	—	—	—	—	114	—	—	197	71.15	73.07	84.61	51.92