Agentic · tools

Hermes Bench leaderboard

Name: benchgap: LLM benchmark scores, measured and estimated
Creator: benchgap
License: https://creativecommons.org/licenses/by-nc/4.0/

As of 2026-10-29, the highest measured score on Hermes Bench is 76.7% by Claude Opus 5.5. 133 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Claude Opus 5.5	76.7%	measured
2	Muse Wonder Full	75.3%	estimated ± 2.9 pp, high confidence
3	Grok 4.7	74.4%	measured
4	Muse Wonder	74.2%	estimated ± 2.9 pp, high confidence
5	Claude Fable 5.1	74.0%	estimated ± 2.9 pp, high confidence
6	GPT-6.1 Sol	73.9%	measured
7	Gemini 4 Carbon	73.8%	estimated ± 0.3 pp, low confidence
8	Gemini 4 Argon	73.8%	estimated ± 0.3 pp, medium confidence
9	GPT-5.6 Sol	73.8%	estimated ± 0.3 pp, medium confidence
10	GPT-6 Astra	73.7%	measured
11	Claude Opus 5	73.6%	estimated ± 2.9 pp, high confidence
12	GPT-5.6 Terra	73.6%	estimated ± 0.3 pp, medium confidence
13	GPT-6 Sol	73.5%	measured
14	Claude Sonnet 5.5	73.5%	measured
15	Kimi K3.1 Pro	73.3%	estimated ± 0.3 pp, medium confidence
16	Muse Spark 1.3	73.0%	estimated ± 2.9 pp, high confidence
17	Qwen3.8 Max Preview	72.8%	estimated ± 2.9 pp, high confidence
18	MiMo-V2.6-Pro	72.8%	estimated ± 0.3 pp, medium confidence
19	Claude Mythos Preview	72.7%	estimated ± 0.3 pp, medium confidence
20	Qwen3.8-Flash-Next	72.3%	estimated ± 2.9 pp, high confidence
21	Grok 4.6	72.1%	estimated ± 2.9 pp, high confidence
22	Ling 3.1 Flash	72.1%	estimated ± 2.9 pp, high confidence
23	Claude Fable 5	72.0%	estimated ± 2.9 pp, high confidence
24	DeepSeek V4.1 Flash 1017	72.0%	estimated ± 0.3 pp, medium confidence
25	GLM-5.3	71.8%	measured
26	DeepSeek V4 Pro 0813	71.7%	estimated ± 2.9 pp, high confidence
27	Kimi K3	71.5%	measured
28	GPT-5.5	71.2%	estimated ± 0.3 pp, medium confidence
29	Claude Haiku 5.5	71.0%	measured
30	GPT-5.6 Luna	70.7%	estimated ± 0.3 pp, medium confidence
31	Qwen3.8 Max	70.5%	measured
32	GPT-6 Luna	70.5%	measured
33	GPT-5.4	69.7%	estimated ± 0.3 pp, low confidence
34	MiMo-V2.6-Flash	69.7%	estimated ± 0.3 pp, low confidence
35	Muse Spark 1.1	69.7%	estimated ± 0.3 pp, low confidence
36	Muse Spark 1.2	69.6%	estimated ± 2.9 pp, high confidence
37	Claude Sonnet 5	69.5%	estimated ± 2.9 pp, high confidence
38	MiniMax M3.1 Flash	69.4%	estimated ± 2.9 pp, high confidence
39	Claude Opus 4.8	69.3%	estimated ± 2.9 pp, high confidence
40	Step 5 Preview	69.2%	measured
41	DeepSeek V4 Flash 0731	68.9%	estimated ± 2.9 pp, high confidence
42	GLM-5.3-Flash	68.9%	measured
43	Ling 3.0 Flash Sante	68.8%	estimated ± 4.8 pp, medium confidence
44	Mistral Large 4	68.7%	estimated ± 2.9 pp, high confidence
45	Qwen3.8-27B	68.6%	estimated ± 2.9 pp, high confidence
46	GPT-6.1 Terra	68.4%	estimated ± 3.4 pp, high confidence
47	Muse Spark Open	68.2%	estimated ± 2.9 pp, high confidence
48	Gemini 3.7 Flash	68.0%	estimated ± 2.9 pp, high confidence
49	Ornith-1.5-397B	68.0%	estimated ± 4.8 pp, medium confidence
50	Grok 4.5	67.9%	estimated ± 2.9 pp, high confidence
51	GLM-5.2	67.6%	estimated ± 2.9 pp, high confidence
52	Sage-A1.5	67.2%	estimated ± 4.8 pp, low confidence
53	Claude Opus 4.7 (Adaptive)	67.2%	estimated ± 2.9 pp, high confidence
54	Gemini 3.5 Flash	67.0%	estimated ± 2.9 pp, high confidence
55	Gemini 3.6 Flash	65.5%	estimated ± 2.9 pp, high confidence
56	dots3-note Preview	65.2%	estimated ± 4.8 pp, low confidence
57	Ling 3.0 Flash Fin	64.5%	estimated ± 8.2 pp, low confidence
58	MiniMax M3	64.4%	estimated ± 2.9 pp, high confidence
59	Beam 1.0 501B-A23B FP8	63.9%	estimated ± 4.8 pp, low confidence
60	Beam 1.0 501B-A23B NVFP4	63.9%	estimated ± 4.8 pp, low confidence
61	Hy3 Preview	63.6%	estimated ± 2.9 pp, high confidence
62	GLM-5	63.3%	estimated ± 4.8 pp, low confidence
63	Apodex 1.1	62.8%	estimated ± 2.9 pp, high confidence
64	Apodex 1.1 Mini	62.8%	estimated ± 2.9 pp, high confidence
65	Quasar 438B	62.6%	estimated ± 2.9 pp, high confidence
66	Beam 1.0 501B-A23B	62.5%	estimated ± 2.9 pp, high confidence
67	Gemini 3.8 Flash	62.2%	measured
68	Nemotron 3.5 Super VL 120B A12B	62.0%	estimated ± 2.9 pp, high confidence
69	Nemotron 3.5 Super VL 120B A12B NVFP4	62.0%	estimated ± 2.9 pp, high confidence
70	Nemotron 3 Ultra	62.0%	estimated ± 2.9 pp, high confidence
71	Ling 3.0 Flash VL	61.9%	estimated ± 2.9 pp, high confidence
72	Agents-A1	60.9%	estimated ± 4.8 pp, low confidence
73	Qwen3.7 Max	60.9%	estimated ± 2.9 pp, high confidence
74	Inkling-Small	60.7%	estimated ± 2.9 pp, high confidence
75	MiMo-V2.5-Pro	60.6%	estimated ± 2.9 pp, high confidence
76	GLM-5.1	60.4%	estimated ± 2.9 pp, high confidence
77	Solar Pro 4	60.1%	estimated ± 2.9 pp, high confidence
78	Grok 4.3	59.3%	estimated ± 2.9 pp, high confidence
79	Inkling	58.8%	estimated ± 2.9 pp, high confidence
80	Hy4 preview	58.6%	measured
81	Hy3	58.5%	estimated ± 2.9 pp, high confidence
82	Kimi K2.6	57.4%	estimated ± 2.9 pp, high confidence
83	Kimi K2.7 Code	57.4%	estimated ± 2.9 pp, high confidence
84	Step 3.7 Flash	56.5%	estimated ± 2.9 pp, high confidence
85	GLM-4.7	56.4%	estimated ± 2.9 pp, high confidence
86	GPT-5.4 mini	56.3%	estimated ± 2.9 pp, high confidence
87	MiniMax M2.7	56.2%	estimated ± 2.9 pp, high confidence
88	Muse Spark	55.7%	estimated ± 2.9 pp, high confidence
89	Qwen3.6 Plus	55.3%	estimated ± 2.9 pp, high confidence
90	Qwen3.6-27B	55.1%	estimated ± 2.9 pp, high confidence
91	Gemini 3.5 Flash-Lite	55.0%	estimated ± 2.9 pp, high confidence
92	Nemotron 3.5 Bolt 74B-A9B	54.8%	estimated ± 2.9 pp, high confidence
93	A.X K2	53.8%	estimated ± 2.9 pp, high confidence
94	Ling 3.0 Flash	53.8%	measured
95	GPT-5.4 nano	53.2%	estimated ± 2.9 pp, high confidence
96	GPT-5 (high)	51.6%	estimated ± 2.9 pp, medium confidence
97	Qwen3.6-35B-A3B	50.1%	estimated ± 2.9 pp, medium confidence
98	Kimi K2.5	46.7%	estimated ± 2.9 pp, medium confidence
99	Kimi K2.5 (Reasoning)	46.7%	estimated ± 2.9 pp, medium confidence
100	GPT-5.1	45.8%	estimated ± 2.9 pp, medium confidence
101	Qwen3.5-122B-A10B	44.9%	estimated ± 2.9 pp, medium confidence
102	K-EXAONE 2.0	44.4%	estimated ± 2.9 pp, medium confidence
103	Gemini 3.1 Pro	43.2%	estimated ± 2.9 pp, medium confidence
104	Muse Glimmer 30B	42.5%	estimated ± 2.9 pp, medium confidence
105	Qwen3.7 Plus	41.0%	estimated ± 2.9 pp, medium confidence
106	Mistral Medium 3.5 128B	40.5%	estimated ± 2.9 pp, medium confidence
107	MiniCPM5-2B	35.4%	estimated ± 2.9 pp, medium confidence
108	Nemotron 3.5 Lightning 30B A3B NVFP4	26.9%	estimated ± 2.9 pp, medium confidence
109	MiMo-V2-Flash	25.4%	estimated ± 2.9 pp, medium confidence
110	Gemma 4 31B	24.0%	estimated ± 2.9 pp, medium confidence
111	GPT-OSS 120B	22.4%	estimated ± 2.9 pp, medium confidence
112	Granite 4.2 30B	16.6%	estimated ± 2.9 pp, medium confidence
113	Gemma 4 26B A4B	14.7%	estimated ± 2.9 pp, medium confidence
114	Ling 3.0 Tiny	13.9%	estimated ± 2.9 pp, medium confidence
115	Command A+	3.7%	estimated ± 2.9 pp, medium confidence
116	Celeris-1	0.0%	estimated ± 2.9 pp, medium confidence
117	Claude 3.5 Haiku	0.0%	estimated ± 2.9 pp, medium confidence
118	DeepSeek V3	0.0%	estimated ± 2.9 pp, medium confidence
119	DeepSeek V3 0324	0.0%	estimated ± 2.9 pp, medium confidence
120	Gemini 2.5 Pro	0.0%	estimated ± 2.9 pp, medium confidence
121	Gemma 3 27B	0.0%	estimated ± 2.9 pp, medium confidence
122	Gemma 4 12B	0.0%	estimated ± 2.9 pp, medium confidence
123	Gemma 4 E2B	0.0%	estimated ± 2.9 pp, medium confidence
124	Gemma 4 E4B	0.0%	estimated ± 2.9 pp, medium confidence
125	GPT-4.1 mini	0.0%	estimated ± 2.9 pp, medium confidence
126	GPT-4.1 nano	0.0%	estimated ± 2.9 pp, medium confidence
127	GPT-4o	0.0%	estimated ± 2.9 pp, medium confidence
128	GPT-4o mini	0.0%	estimated ± 2.9 pp, medium confidence
129	GPT-OSS 20B	0.0%	estimated ± 2.9 pp, medium confidence
130	Granite 4.2 3B	0.0%	estimated ± 2.9 pp, medium confidence
131	Granite 4.2 8B	0.0%	estimated ± 2.9 pp, medium confidence
132	K-Exaone	0.0%	estimated ± 2.9 pp, medium confidence
133	LFM2.5-2.6B	0.0%	estimated ± 2.9 pp, medium confidence
134	Ling 2.6 Flash	0.0%	estimated ± 2.9 pp, medium confidence
135	Llama 3.3 70B	0.0%	estimated ± 2.9 pp, medium confidence
136	Llama 4 Maverick	0.0%	estimated ± 2.9 pp, medium confidence
137	Llama 4 Scout	0.0%	estimated ± 2.9 pp, medium confidence
138	Mercury 2.5	0.0%	estimated ± 2.9 pp, medium confidence
139	Mistral Large 3	0.0%	estimated ± 2.9 pp, medium confidence
140	Mistral Small 3.1	0.0%	estimated ± 2.9 pp, medium confidence
141	Mistral Small 4	0.0%	estimated ± 2.9 pp, medium confidence
142	Mistral Small 4 (Reasoning)	0.0%	estimated ± 2.9 pp, medium confidence
143	Nemotron 3 Nano 30B	0.0%	estimated ± 2.9 pp, medium confidence
144	Nemotron 3 Nano Omni 30B A3B	0.0%	estimated ± 2.9 pp, medium confidence
145	North Mini Code	0.0%	estimated ± 2.9 pp, medium confidence
146	Solar Pro 3	0.0%	estimated ± 2.9 pp, medium confidence
147	Trinity-Large-Preview	0.0%	estimated ± 2.9 pp, medium confidence
148	Trinity-Large-Thinking	0.0%	estimated ± 2.9 pp, medium confidence
149	Ultravox v0.6 Llama 3.3 70B	0.0%	estimated ± 2.9 pp, medium confidence

Hermes Bench leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General