Question Answering on HotpotQA τ = 96k benchmark leaderboard · SOTA2 Research

Question Answering on HotpotQA τ = 96k

86.9E2E Wall Time (s)gpt-oss-20B

Evaluation Results

Method	Links
gpt-oss-20B2026.05		86.9	1	145.2	1	1,493	1	11,127	1	5.34	1
gpt-oss-20B2026.05		100.7	0.86	147.1	1.01	3,098	2.07	11,704	1.05	5.59	0.95
gpt-oss-20B2026.05		101.8	0.85	168.1	1.16	5,401	3.62	11,710	1.05	3.39	1.58
Llama-3.1-8B2026.05		104.6	1.44	62.8	1.38	1,151	0.65	5,424	1.07	17.89	0.8
gpt-oss-20B2026.05		105.1	0.83	205.3	1.41	10,017	6.71	11,554	1.04	2.12	2.52
Llama-3.1-8B2026.05		105.8	1.43	71	1.56	2,199	1.24	5,313	1.05	10.43	1.37
gpt-oss-20B2026.05		108.3	0.8	253.8	1.75	16,180	10.84	11,303	1.02	1.53	3.48
Llama-3.1-8B2026.05		119.4	1.26	83.8	1.84	4,575	2.58	5,433	1.07	6.57	2.18
Llama-3.1-8B2026.05		131.7	1.14	45.2	0.99	681	0.38	5,272	1.04	32.96	0.43
Llama-3.1-8B2026.05		150.8	1	45.5	1	1,776	1	5,082	1	14.31	1
gpt-oss-120B2026.05		214.5	1	51	1	1,462	1	9,480	1	7.28	1
gpt-oss-120B2026.05		217.7	0.99	90	1.76	9,894	6.77	9,701	1.02	3.58	2.04
gpt-oss-120B2026.05		233.1	0.92	109	2.14	16,486	11.28	8,913	0.94	2.51	2.91
Llama-3.3-70B2026.05		266.6	1.4	52.5	1.35	8,360	0.97	5,645	0.94	10.62	2.13
Llama-3.3-70B2026.05		268.5	1.39	46.4	1.19	6,823	0.79	5,647	0.94	13.29	1.7
gpt-oss-120B2026.05		270.3	0.79	59.7	1.17	6,994	4.78	9,136	0.96	5.22	1.39
gpt-oss-120B2026.05		276	0.78	49.8	0.98	4,116	2.81	9,615	1.01	7.64	0.95
Llama-3.3-70B2026.05		289.7	1.29	39.6	1.01	5,317	0.62	6,145	1.03	19.79	1.14
Llama-3.3-70B2026.05		292.4	1.28	64.1	1.64	12,303	1.43	6,427	1.07	8.09	2.8
Llama-3.3-70B2026.05		373.5	1	39	1	8,582	1	5,994	1	22.62	1