Agentic · tools

MCP Atlas leaderboard

Name: benchgap: LLM benchmark scores, measured and estimated
Creator: benchgap

As of 2026-10-07, the highest measured score on MCP Atlas is 88.1% by Muse Spark 1.1. 151 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Grok 4.6	100.0%	estimated ± 0.7 pp, low confidence
2	Gemini 4 Argon	89.2%	estimated ± 2.7 pp, low confidence
3	Muse Spark 1.1	88.1%	measured
4	Claude Sonnet 5.5	88.0%	estimated ± 2.7 pp, low confidence
5	Claude Opus 5.5	87.5%	estimated ± 2.7 pp, low confidence
6	DeepSeek V4.1 Flash	87.4%	estimated ± 2.7 pp, low confidence
7	GPT-6 Astra	87.3%	estimated ± 2.7 pp, low confidence
8	Grok 4.7	86.7%	estimated ± 2.7 pp, low confidence
9	GPT-6.1 Sol	86.5%	estimated ± 2.7 pp, low confidence
10	GPT-5.6 Sol	86.3%	estimated ± 3.5 pp, low confidence
11	GLM-5.3	85.9%	estimated ± 2.7 pp, low confidence
12	GPT-6 Sol	85.8%	estimated ± 2.7 pp, low confidence
13	Claude Opus 5	85.8%	measured
14	Step 5 Preview	85.6%	measured
15	Apodex 1.1	85.6%	estimated ± 0.7 pp, low confidence
16	GLM-5.3-Flash	85.6%	estimated ± 2.7 pp, low confidence
17	Gemini 3.8 Flash	85.5%	estimated ± 2.7 pp, low confidence
18	Mistral Large 4	85.5%	estimated ± 2.7 pp, low confidence
19	Claude Fable 5.1	85.3%	estimated ± 2.7 pp, low confidence
20	MiMo-V2.6-Pro	85.2%	estimated ± 2.7 pp, low confidence
21	Muse Spark 1.3	85.0%	estimated ± 2.7 pp, medium confidence
22	Gemini 3.7 Flash	84.8%	estimated ± 6.2 pp, low confidence
23	Kimi K3	84.2%	measured
24	Ling 3.1 Flash	84.2%	estimated ± 11.1 pp, low confidence
25	GPT-6 Luna	84.0%	estimated ± 2.7 pp, medium confidence
26	Hy4 preview	83.7%	measured
27	Gemini 3.5 Flash	83.6%	measured
28	Qwen3.8-27B	82.9%	estimated ± 2.7 pp, medium confidence
29	GPT-5.6 Terra	82.6%	estimated ± 3.5 pp, medium confidence
30	Claude Opus 4.8	82.2%	measured
31	Claude Fable 5	82.1%	estimated ± 5.5 pp, low confidence
32	Holo3-35B-A3B	81.3%	estimated ± 11.5 pp, low confidence
33	MiMo-V2.6-Flash	81.0%	estimated ± 7.8 pp, medium confidence
34	Qwen3.8-Flash-Next	80.9%	estimated ± 7.8 pp, medium confidence
35	Claude Haiku 5.5	80.2%	estimated ± 2.7 pp, medium confidence
36	Ornith-1.5-397B	80.0%	measured
37	Claude Opus 4.7	79.6%	estimated ± 10.4 pp, low confidence
38	Inkling-Small	79.6%	measured
39	Claude Sonnet 5	79.5%	estimated ± 7.3 pp, medium confidence
40	Fugu Cyber	79.0%	estimated ± 11.4 pp, low confidence
41	Gemini 3.8 Flash Cyber	78.9%	estimated ± 11.4 pp, low confidence
42	Qwen3.8 Max	78.9%	estimated ± 7.3 pp, medium confidence
43	Beam	78.7%	measured
44	GPT-5.6 Luna	78.7%	estimated ± 7.8 pp, low confidence
45	Grok 4.5	78.7%	estimated ± 7.8 pp, low confidence
46	Muse Spark 1.2	78.7%	estimated ± 7.8 pp, low confidence
47	Quasar 438B	78.7%	estimated ± 7.8 pp, low confidence
48	Qwen3.8 Max Preview	78.7%	estimated ± 7.8 pp, low confidence
49	Gemini 3.6 Flash	78.7%	estimated ± 7.8 pp, low confidence
50	Gemini 3.5 Flash Cyber	78.5%	estimated ± 11.4 pp, low confidence
51	Claude Mythos Preview	78.5%	estimated ± 11.4 pp, low confidence
52	Hy3	78.3%	estimated ± 7.8 pp, low confidence
53	Hy3 Preview	78.3%	estimated ± 7.8 pp, low confidence
54	GPT-5.5 Pro	78.2%	estimated ± 10.0 pp, low confidence
55	Atria Dawn Preview	78.1%	estimated ± 8.2 pp, medium confidence
56	GPT-5.4 Pro	77.8%	estimated ± 10.0 pp, low confidence
57	K-EXAONE 2.0	77.5%	estimated ± 11.2 pp, low confidence
58	Claude Opus 4.7 (Adaptive)	77.3%	measured
59	Claude Mythos 5	77.1%	estimated ± 10.0 pp, low confidence
60	dots3-note Preview	77.0%	estimated ± 0.7 pp, medium confidence
61	Ornith-1.0-397B	77.0%	estimated ± 11.2 pp, low confidence
62	GLM-5.2	76.8%	measured
63	Holo3-122B-A10B	76.6%	estimated ± 11.5 pp, low confidence
64	Qwen3.7 Max	76.4%	measured
65	Kimi K2.7 Code	76.0%	measured
66	Muse Glimmer 30B	75.5%	measured
67	GPT-5.5	75.3%	measured
68	Ling 3.0 Flash Fin	75.1%	estimated ± 0.7 pp, medium confidence
69	UI-Mate-27B	74.3%	estimated ± 11.5 pp, low confidence
70	MiniMax M3	74.2%	measured
71	Inkling	74.1%	measured
72	Agents-A1	73.8%	estimated ± 7.3 pp, medium confidence
73	DeepSeek V4 Pro 0813	73.6%	measured
74	Step 3.7 Flash	73.5%	estimated ± 7.3 pp, medium confidence
75	Apodex 1.1 Mini	73.3%	estimated ± 0.7 pp, medium confidence
76	Nemotron 3 Ultra	73.3%	estimated ± 2.7 pp, low confidence
77	Qwen3.7 Plus	73.2%	measured
78	Claude Opus 4.6 (Adaptive)	72.3%	estimated ± 9.4 pp, medium confidence
79	GLM-5.1	71.8%	measured
80	Qwen 3.6 Max (preview)	71.1%	estimated ± 12.3 pp, low confidence
81	GPT-5.4	70.6%	measured
82	Ling 3.0 Flash VL	70.3%	estimated ± 11.1 pp, low confidence
83	Ornith-1.5-35B-A3B	70.2%	measured
84	Ornith-1.0-35B	69.3%	estimated ± 11.2 pp, low confidence
85	DeepSeek V4 Flash 0731	69.0%	measured
86	MiMo-V2.5-Pro	68.5%	estimated ± 7.8 pp, low confidence
87	Laguna S 2.1	67.4%	estimated ± 7.8 pp, medium confidence
88	Ling 3.0 Flash	65.5%	measured
89	Muse Spark	65.4%	estimated ± 11.1 pp, low confidence
90	Ling 3.0 Flash FP8	65.0%	estimated ± 7.8 pp, low confidence
91	Qwen3.6-27B	64.7%	estimated ± 7.8 pp, low confidence
92	Grok 4.3	64.6%	estimated ± 7.8 pp, low confidence
93	MiniMax M2.7	64.6%	estimated ± 7.8 pp, low confidence
94	Gemini 3.5 Flash-Lite	64.6%	estimated ± 7.8 pp, low confidence
95	Celeris-1	64.6%	estimated ± 7.8 pp, low confidence
96	Command A+	64.6%	estimated ± 7.8 pp, low confidence
97	DeepSeek V3	64.6%	estimated ± 7.8 pp, low confidence
98	Gemini 2.5 Pro	64.6%	estimated ± 7.8 pp, low confidence
99	Gemini 3.1 Pro	64.6%	estimated ± 7.8 pp, low confidence
100	Gemma 3 27B	64.6%	estimated ± 7.8 pp, low confidence
101	Gemma 4 31B	64.6%	estimated ± 7.8 pp, low confidence
102	GPT-OSS 120B	64.6%	estimated ± 7.8 pp, low confidence
103	GPT-OSS 20B	64.6%	estimated ± 7.8 pp, low confidence
104	Granite 4.2 8B	64.6%	estimated ± 7.8 pp, low confidence
105	Llama 4 Maverick	64.6%	estimated ± 7.8 pp, low confidence
106	Llama 4 Scout	64.6%	estimated ± 7.8 pp, low confidence
107	Mistral Large 3	64.6%	estimated ± 7.8 pp, low confidence
108	Mistral Medium 3.5 128B	64.6%	estimated ± 7.8 pp, low confidence
109	Mistral Small 4	64.6%	estimated ± 7.8 pp, low confidence
110	Mistral Small 4 (Reasoning)	64.6%	estimated ± 7.8 pp, low confidence
111	Nemotron 3.5 Lightning 30B A3B NVFP4	64.6%	estimated ± 7.8 pp, low confidence
112	Nemotron 3 Nano 30B	64.6%	estimated ± 7.8 pp, low confidence
113	Nemotron 3 Super 100B	64.6%	estimated ± 7.8 pp, low confidence
114	Qwen3.5-122B-A10B	64.6%	estimated ± 7.8 pp, low confidence
115	Trinity-Large-Preview	64.6%	estimated ± 7.8 pp, low confidence
116	Trinity-Large-Thinking	64.6%	estimated ± 7.8 pp, low confidence
117	Agents-A1-4B	64.5%	estimated ± 10.0 pp, low confidence
118	A.X K2	64.2%	estimated ± 11.1 pp, low confidence
119	Qwen3.6-35B-A3B	62.8%	measured
120	UI-Mate-9B	62.4%	estimated ± 11.5 pp, low confidence
121	Claude Sonnet 4.6	62.2%	estimated ± 8.2 pp, medium confidence
122	Claude Opus 4.6	61.9%	estimated ± 8.2 pp, medium confidence
123	Solar Pro 4	61.4%	measured
124	Qwen3.5-27B	60.6%	estimated ± 10.0 pp, low confidence
125	Qwen3.5-35B-A3B	60.6%	estimated ± 10.0 pp, low confidence
126	GPT-5.1	60.2%	estimated ± 11.1 pp, low confidence
127	GPT-5.2	59.1%	estimated ± 8.2 pp, medium confidence
128	Ornith-1.0-9B	58.5%	estimated ± 11.2 pp, low confidence
129	GPT-5.3 Codex	58.4%	estimated ± 8.2 pp, medium confidence
130	Solar Open 2	58.2%	measured
131	GPT-5.4 mini	57.7%	measured
132	MiMo-V2.5	57.0%	estimated ± 11.2 pp, low confidence
133	MiniCPM5-2B	56.5%	estimated ± 11.1 pp, low confidence
134	GPT-5.4 nano	56.1%	measured
135	Kimi K2.6	55.9%	measured
136	Ornith-1.5-9B	54.2%	measured
137	MiMo-V2-Flash	54.2%	estimated ± 11.1 pp, low confidence
138	GLM-4.7	54.1%	estimated ± 10.0 pp, low confidence
139	Granite 4.2 30B	52.5%	estimated ± 11.1 pp, low confidence
140	Gemma 4 26B A4B	52.2%	estimated ± 11.1 pp, low confidence
141	Ling 3.0 Tiny	52.1%	estimated ± 11.1 pp, low confidence
142	Claude Sonnet 4.5	51.2%	estimated ± 8.2 pp, medium confidence
143	DeepSeek V3 0324	50.2%	estimated ± 11.1 pp, low confidence
144	Gemma 4 12B	50.2%	estimated ± 11.1 pp, low confidence
145	Gemma 4 E2B	50.2%	estimated ± 11.1 pp, low confidence
146	Gemma 4 E4B	50.2%	estimated ± 11.1 pp, low confidence
147	GPT-4.1 mini	50.2%	estimated ± 11.1 pp, low confidence
148	GPT-4.1 nano	50.2%	estimated ± 11.1 pp, low confidence
149	GPT-4o	50.2%	estimated ± 11.1 pp, low confidence
150	GPT-4o mini	50.2%	estimated ± 11.1 pp, low confidence
151	Granite 4.2 3B	50.2%	estimated ± 11.1 pp, low confidence
152	K-Exaone	50.2%	estimated ± 11.1 pp, low confidence
153	LFM2.5-2.6B	50.2%	estimated ± 11.1 pp, low confidence
154	Ling 2.6 Flash	50.2%	estimated ± 11.1 pp, low confidence
155	Mercury 2.5	50.2%	estimated ± 11.1 pp, low confidence
156	Nemotron 3 Nano Omni 30B A3B	50.2%	estimated ± 11.1 pp, low confidence
157	North Mini Code	50.2%	estimated ± 11.1 pp, low confidence
158	Solar Pro 3	50.2%	estimated ± 11.1 pp, low confidence
159	Ultravox v0.6 Llama 3.3 70B	50.2%	estimated ± 11.1 pp, low confidence
160	GPT-5.1-Codex	49.4%	estimated ± 8.2 pp, medium confidence
161	GPT-5.2-Codex	49.2%	estimated ± 8.2 pp, medium confidence
162	Qwen3.6 Plus	48.2%	measured
163	MiMo-V2-Pro	47.3%	estimated ± 11.2 pp, low confidence
164	LLaDA2.2-flash	46.2%	measured
165	Qwen3.5 397B	46.1%	measured
166	LLaDA2.2-mini	45.8%	estimated ± 11.2 pp, low confidence
167	LongCat-Flash-Lite-Sparse	45.6%	measured
168	Claude 4.1 Opus	44.3%	estimated ± 8.2 pp, medium confidence
169	Gemini 3.1 Flash-Lite	43.5%	estimated ± 14.8 pp, low confidence
170	GPT-4.1	43.5%	estimated ± 14.8 pp, low confidence
171	Grok 4	43.5%	estimated ± 14.8 pp, low confidence
172	Grok 4.1 Fast	43.5%	estimated ± 14.8 pp, low confidence
173	Grok 4.20	43.5%	estimated ± 14.8 pp, low confidence
174	Grok Build 0.1	43.5%	estimated ± 14.8 pp, low confidence
175	Qwen3 Max	43.5%	estimated ± 14.8 pp, low confidence
176	GLM-5-Turbo	42.6%	estimated ± 11.2 pp, low confidence
177	Claude Opus 4.5	42.3%	measured
178	Qwen3.5 Plus	40.3%	estimated ± 8.2 pp, medium confidence
179	Claude 4 Sonnet	40.1%	estimated ± 8.2 pp, medium confidence
180	Grok 4.1	38.6%	estimated ± 13.5 pp, low confidence
181	GLM-5V-Turbo	37.8%	estimated ± 11.2 pp, low confidence
182	Claude Haiku 4.5	37.3%	estimated ± 8.2 pp, medium confidence
183	Kimi K2.5 (Reasoning)	33.9%	estimated ± 9.4 pp, medium confidence
184	Gemini 3 Flash	31.8%	estimated ± 8.2 pp, medium confidence
185	Gemini 3 Pro	31.8%	estimated ± 8.2 pp, medium confidence
186	GLM-5	31.1%	measured
187	Kimi K2.5	29.5%	measured
188	GPT-5 (high)	28.4%	estimated ± 8.2 pp, low confidence
189	MiMo-V2-Omni	18.3%	estimated ± 11.2 pp, low confidence
190	DeepSeek V3.2	10.1%	estimated ± 11.2 pp, low confidence

MCP Atlas leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General