Plan Generation on 60-task evaluation suite 10 domains benchmark leaderboard · SOTA2 Research

Plan Generation on 60-task evaluation suite 10 domains

89Code PQSPII Redaction

Evaluation Results

Method	Links
PII Redaction2026.03		89	40	71	95	89	91	100	90	83	95	86.1	84.8	83.5	16.5
PII Redaction2026.03		85	80	75	88	93	93	89	88	88	100	88.7	88.6	83.5	16.5
Raw Context2026.03		81	40	67	90	95	88	87	78	91	100	83.2	83.3	0	100
PlanTwin2026.03		81	75	64	72	98	80	87	67	93	83	81	76.5	100	61.5
PlanTwin2026.03		79	80	68	72	100	74	79	67	90	91	80.8	77.6	100	61.7
Raw Context2026.03		79	65	67	76	83	77	87	78	95	89	80.2	76.1	0	100
Raw Context2026.03		77	77	67	81	85	80	93	84	93	98	84.3	83.1	0	100
Raw Context2026.03		77	40	60	89	90	87	100	90	82	98	83.4	84.5	0	100
PII Redaction2026.03		77	40	67	94	95	95	100	93	78	94	85.5	87.2	83.5	16.5
PlanTwin2026.03		75	69	67	83	100	70	86	73	72	81	79.2	79.9	100	59
PlanTwin2026.03		71	68	57	72	93	48	68	70	66	77	68.9	60.4	100	58.2
PII Redaction2026.03		65	75	73	77	78	79	87	75	91	83	78.8	74.7	83.5	16.5
Local (Gemma-3-1B)2026.03		55	86	50	54	60	57	55	55	55	44	56.7	55.9	100	0
Heuristic (Oracle)2026.03		53	50	50	78	90	70	59	54	43	84	65.3	78.2	100	55.8
Local (Qwen 3.5-4B)2026.03		27	40	50	46	58	50	52	70	54	30	48.4	65.6	100	0
Local (SmolLM2-360M)‡2026.03		10	50	50	12	38	50	17	33	17	17	28.3	56.7	100	0
Local (Qwen 3.5-0.8B)‡2026.03		10	50	50	12	38	50	17	33	17	17	28.3	56.7	100	0
Local (Qwen 3.5-2B)2026.03		10	50	50	12	38	50	17	33	17	17	28.3	56.7	100	0
Local (Qwen 3.5-9B)2026.03		10	50	86	12	64	93	73	50	45	60	53.1	64.2	100	0